[mlir] Replace StrEnumAttr -> EnumAttr in core dialects

[Mogball]

Summary

Removes uses of StrEnumAttr in core dialects

[mehdi_amini]

Nice! I like the removal of string comparison to use enums instead :)

[Mogball]

Indeed. Surprisingly, some dialects required very little work to change (StrEnumAttr -> EnumAttr and fix the BUILD file and that's it).

[jpienaar]

mlir/include/mlir/Dialect/GPU/GPUOps.td

40

Unrelated here, but we have namespace here and dialect specified below (which has a namespace), feels a bit weird

[Mogball]

mlir/include/mlir/Dialect/GPU/GPUOps.td

40

Yeah. That's because enums aren't tied to any particular dialect. I would like to redo enum gen entirely once all enums are moved over to AttrDef.

[Mogball]

Fix flang and remove StrEnumAttr from OMP/ACC directives gen

[Mogball]

rebase

[rriddle]

LGTM, nice cleanup!

mlir/include/mlir/Dialect/GPU/CMakeLists.txt

35

Is the m(docm) here an accident?

mlir/include/mlir/Dialect/GPU/GPUOps.td

680

Please wrap this at 80 characters.

mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp

556

mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp

189

[Mogball]

Thanks for the review

[Mogball]

review comments

mlir/include/mlir/Dialect/GPU/CMakeLists.txt

 mlir_tablegen(Passes.capi.cpp.inc -gen-pass-capi-impl --prefix GPU)
 add_public_tablegen_target(MLIRGPUPassIncGen)
 
 set(LLVM_TARGET_DEFINITIONS GPUOps.td)
 mlir_tablegen(GPUOpsEnums.h.inc -gen-enum-decls)
 mlir_tablegen(GPUOpsEnums.cpp.inc -gen-enum-defs)
 add_public_tablegen_target(MLIRGPUOpsEnumsGen)
 
-add_mlir_doc(Passes GPUPasses ./ -gen-pass-doc)
+set(LLVM_TARGET_DEFINITIONS GPUOps.td)
+mlir_tablegen(GPUOpsAttributes.h.inc -gen-attrdef-decls -attrdefs-dialect=gpu)
+mlir_tablegen(GPUOpsAttributes.cpp.inc -gen-attrdef-defs -attrdefs-dialect=gpu)
+add_public_tablegen_target(MLIRGPUOpsAttributesIncGen)
+
+add_mlir_doc(Passes GPUPasses ./ -gen-pass-docm)

mlir/include/mlir/Dialect/GPU/GPUBase.td

let extraClassDeclaration = [{
     static unsigned getWorkgroupAddressSpace() { return 3; }
 
     /// Returns the numeric value used to identify the private memory address
     /// space.
     static unsigned getPrivateAddressSpace() { return 5; }
   }];
 
   let dependentDialects = ["arith::ArithmeticDialect"];
+  let useDefaultAttributePrinterParser = 1;
 }
 
 def GPU_AsyncToken : DialectType<
   GPU_Dialect, CPred<"$_self.isa<::mlir::gpu::AsyncTokenType>()">, "async token type">,
              BuildableType<"mlir::gpu::AsyncTokenType::get($_builder.getContext())">;
 
 // Predicat to check if type is gpu::MMAMatrixType.
 def IsMMAMatrixTypePred : CPred<"$_self.isa<::mlir::gpu::MMAMatrixType>()">;

mlir/include/mlir/Dialect/GPU/GPUDialect.h

 } // namespace mlir
 
 #include "mlir/Dialect/GPU/GPUOpsEnums.h.inc"
 
 #include "mlir/Dialect/GPU/GPUOpsDialect.h.inc"
 
 #include "mlir/Dialect/GPU/GPUOpInterfaces.h.inc"
 
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/GPU/GPUOpsAttributes.h.inc"
+
 #define GET_OP_CLASSES
 #include "mlir/Dialect/GPU/GPUOps.h.inc"
 
 #endif // MLIR_DIALECT_GPU_GPUDIALECT_H

mlir/include/mlir/Dialect/GPU/GPUOps.td

 //===----------------------------------------------------------------------===//
 
 #ifndef GPU_OPS
 #define GPU_OPS
 
 include "mlir/Dialect/DLTI/DLTIBase.td"
 include "mlir/Dialect/GPU/GPUBase.td"
 include "mlir/Dialect/LLVMIR/LLVMOpBase.td"
+include "mlir/IR/EnumAttr.td"
 include "mlir/IR/SymbolInterfaces.td"
 include "mlir/Interfaces/DataLayoutInterfaces.td"
 include "mlir/Interfaces/InferTypeOpInterface.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
 
 //===----------------------------------------------------------------------===//
 // GPU Dialect operations.
 //===----------------------------------------------------------------------===//
 
 class GPU_Op<string mnemonic, list<OpTrait> traits = []> :
     Op<GPU_Dialect, mnemonic, traits>;
 
+def GPU_Dimension : I32EnumAttr<"Dimension",
+    "a dimension, either 'x', 'y', or 'z'",
+    [
+      I32EnumAttrCase<"x", 0>,
+      I32EnumAttrCase<"y", 1>,
+      I32EnumAttrCase<"z", 2>
+    ]>{
+  let genSpecializedAttr = 0;
+  let cppNamespace = "::mlir::gpu";
+}
+def GPU_DimensionAttr : EnumAttr<GPU_Dialect, GPU_Dimension, "dim">;
+
 class GPU_IndexOp<string mnemonic, list<OpTrait> traits = []> :
     GPU_Op<mnemonic, !listconcat(traits, [NoSideEffect])>,
-    Arguments<(ins StrAttr:$dimension)>, Results<(outs Index)> {
-  let verifier = [{ return ::verifyIndexOp(*this); }];
+    Arguments<(ins GPU_DimensionAttr:$dimension)>, Results<(outs Index)> {
+  let assemblyFormat = "$dimension attr-dict";
 }
 
 def GPU_BlockDimOp : GPU_IndexOp<"block_dim"> {
   let description = [{
     Returns the number of threads in the thread block (aka the block size) along
     the x, y, or z `dimension`.
 
     Example:
 
     ```mlir
-    %bDimX = "gpu.block_dim"() {dimension = "x"} : () -> (index)
+    %bDimX = gpu.block_dim x
     ```
   }];
 }
 def GPU_BlockIdOp : GPU_IndexOp<"block_id"> {
   let description = [{
     Returns the block id, i.e. the index of the current block within the grid
     along the x, y, or z `dimension`.
 
     Example:
 
     ```mlir
-    %bIdY = "gpu.block_id"() {dimension = "y"} : () -> (index)
+    %bIdY = gpu.block_id y
     ```
   }];
 }
 def GPU_GridDimOp : GPU_IndexOp<"grid_dim"> {
   let description = [{
     Returns the number of thread blocks in the grid along the x, y, or z
     `dimension`.
 
     Example:
 
     ```mlir
-    %gDimZ = "gpu.grid_dim"() {dimension = "z"} : () -> (index)
+    %gDimZ = gpu.grid_dim z
     ```
   }];
 }
 def GPU_ThreadIdOp : GPU_IndexOp<"thread_id"> {
   let description = [{
     Returns the thread id, i.e. the index of the current thread within the block
     along the x, y, or z `dimension`.
 
     Example:
 
     ```mlir
-    %tIdX = "gpu.thread_id"() {dimension = "x"} : () -> (index)
+    %tIdX = gpu.thread_id x
     ```
   }];
 }
 
 def GPU_SubgroupIdOp : GPU_Op<"subgroup_id", [NoSideEffect]>,
     Arguments<(ins)>, Results<(outs Index:$result)> {
   let description = [{
     Returns the subgroup id, i.e. the index of the current subgroup within the
module attributes {gpu.container_module} {
       // This module creates a separate compilation unit for the GPU compiler.
       gpu.module @kernels {
         func @kernel_1(%arg0 : f32, %arg1 : memref<?xf32, 1>)
             attributes { nvvm.kernel = true } {
 
           // Operations that produce block/thread IDs and dimensions are
           // injected when outlining the `gpu.launch` body to a function called
           // by `gpu.launch_func`.
-          %tIdX = "gpu.thread_id"() {dimension = "x"} : () -> (index)
-          %tIdY = "gpu.thread_id"() {dimension = "y"} : () -> (index)
-          %tIdZ = "gpu.thread_id"() {dimension = "z"} : () -> (index)
-
-          %bDimX = "gpu.block_dim"() {dimension = "x"} : () -> (index)
-          %bDimY = "gpu.block_dim"() {dimension = "y"} : () -> (index)
-          %bDimZ = "gpu.block_dim"() {dimension = "z"} : () -> (index)
-
-          %bIdX = "gpu.block_id"() {dimension = "x"} : () -> (index)
-          %bIdY = "gpu.block_id"() {dimension = "y"} : () -> (index)
-          %bIdZ = "gpu.block_id"() {dimension = "z"} : () -> (index)
-
-          %gDimX = "gpu.grid_dim"() {dimension = "x"} : () -> (index)
-          %gDimY = "gpu.grid_dim"() {dimension = "y"} : () -> (index)
-          %gDimZ = "gpu.grid_dim"() {dimension = "z"} : () -> (index)
+          %tIdX = gpu.thread_id x
+          %tIdY = gpu.thread_id y
+          %tIdZ = gpu.thread_id z
+
+          %bDimX = gpu.block_dim x
+          %bDimY = gpu.block_dim y
+          %bDimZ = gpu.block_dim z
+
+          %bIdX = gpu.block_id x
+          %bIdY = gpu.block_id y
+          %bIdZ = gpu.block_id z
+
+          %gDimX = gpu.grid_dim x
+          %gDimY = gpu.grid_dim y
+          %gDimZ = gpu.grid_dim z
 
           "some_op"(%bx, %tx) : (index, index) -> ()
           %42 = load %arg1[%bx] : memref<?xf32, 1>
         }
       }
 
       %t0 = gpu.wait async
       gpu.launch_func
let description = [{
 
     ```mlir
     gpu.yield %f0, %f1 : f32, f32
     ```
   }];
 }
 
 // add, mul mirror the XLA ComparisonDirection enum.
-def GPU_AllReduceOpAdd : StrEnumAttrCase<"ADD", -1, "add">;
-def GPU_AllReduceOpAnd : StrEnumAttrCase<"AND", -1, "and">;
-def GPU_AllReduceOpMax : StrEnumAttrCase<"MAX", -1, "max">;
-def GPU_AllReduceOpMin : StrEnumAttrCase<"MIN", -1, "min">;
-def GPU_AllReduceOpMul : StrEnumAttrCase<"MUL", -1, "mul">;
-def GPU_AllReduceOpOr : StrEnumAttrCase<"OR", -1, "or">;
-def GPU_AllReduceOpXor : StrEnumAttrCase<"XOR", -1, "xor">;
+def GPU_AllReduceOpAdd : I32EnumAttrCase<"ADD", 0, "add">;
+def GPU_AllReduceOpAnd : I32EnumAttrCase<"AND", 1, "and">;
+def GPU_AllReduceOpMax : I32EnumAttrCase<"MAX", 2, "max">;
+def GPU_AllReduceOpMin : I32EnumAttrCase<"MIN", 3, "min">;
+def GPU_AllReduceOpMul : I32EnumAttrCase<"MUL", 4, "mul">;
+def GPU_AllReduceOpOr  : I32EnumAttrCase<"OR",  5, "or">;
+def GPU_AllReduceOpXor : I32EnumAttrCase<"XOR", 6, "xor">;
 
-def GPU_AllReduceOperationAttr : StrEnumAttr<"AllReduceOperationAttr",
+def GPU_AllReduceOperation : I32EnumAttr<"AllReduceOperation",
     "built-in reduction operations supported by gpu.allreduce.",
     [
       GPU_AllReduceOpAdd,
       GPU_AllReduceOpAnd,
       GPU_AllReduceOpMax,
       GPU_AllReduceOpMin,
       GPU_AllReduceOpMul,
       GPU_AllReduceOpOr,
       GPU_AllReduceOpXor
     ]>{
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::gpu";
 }
+def GPU_AllReduceOperationAttr : EnumAttr<GPU_Dialect, GPU_AllReduceOperation,
+                                          "all_reduce_op">;
 
 def GPU_AllReduceOp : GPU_Op<"all_reduce",
     [SameOperandsAndResultType, IsolatedFromAbove]>,
     Arguments<(ins AnyType:$value,
                OptionalAttr<GPU_AllReduceOperationAttr>:$op)>,
     Results<(outs AnyType)> {
   let summary = "Reduce values among workgroup.";
   let description = [{
     The `all_reduce` op reduces the value of every work item across a local
     workgroup. The result is equal for all work items of a workgroup.
 
     For example, both
 
     ```mlir
-    %1 = "gpu.all_reduce"(%0) ({}) { op = "add" } : (f32) -> (f32)
-    %2 = "gpu.all_reduce"(%0) ({
+    %1 = gpu.all_reduce add %0 {} : (f32) -> (f32)
+    %2 = gpu.all_reduce %0 {
     ^bb(%lhs : f32, %rhs : f32):
       %sum = arith.addf %lhs, %rhs : f32
       "gpu.yield"(%sum) : (f32) -> ()
-    }) : (f32) -> (f32)
+    } : (f32) -> (f32)
     ```
 
     compute the sum of each work item's %0 value. The first version specifies
     the accumulation as operation, whereas the second version specifies the
     accumulation as code region. The accumulation operation must be one of:
     `add`, `and`, `max`, `min`, `mul`, `or`, `xor`.
 
     Either none or all work items of a workgroup need to execute this op
     in convergence.
   }];
   let regions = (region AnyRegion:$body);
   let verifier = [{ return ::verifyAllReduce(*this); }];
+  let assemblyFormat = "custom<AllReduceOperation>($op) $value $body attr-dict `:` functional-type(operands, results)";
 }
 
-def GPU_ShuffleOpXor : StrEnumAttrCase<"XOR", -1, "xor">;
-def GPU_ShuffleOpDown : StrEnumAttrCase<"DOWN", -1, "down">;
-def GPU_ShuffleOpUp : StrEnumAttrCase<"UP", -1, "up">;
-def GPU_ShuffleOpIdx : StrEnumAttrCase<"IDX", -1, "idx">;
+def GPU_ShuffleOpXor  : I32EnumAttrCase<"XOR",  0, "xor">;
+def GPU_ShuffleOpDown : I32EnumAttrCase<"DOWN", 1, "down">;
+def GPU_ShuffleOpUp   : I32EnumAttrCase<"UP",   2, "up">;
+def GPU_ShuffleOpIdx  : I32EnumAttrCase<"IDX",  3, "idx">;
 
-def GPU_ShuffleModeAttr : StrEnumAttr<"ShuffleModeAttr",
+def GPU_ShuffleMode : I32EnumAttr<"ShuffleMode",
     "Indexing modes supported by gpu.shuffle.",
     [
       GPU_ShuffleOpXor, GPU_ShuffleOpUp, GPU_ShuffleOpDown, GPU_ShuffleOpIdx,
-    ]>{
+    ]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::gpu";
-  let storageType = "mlir::StringAttr";
-  let returnType = "::mlir::gpu::ShuffleModeAttr";
-  let convertFromStorage =
-    "*symbolizeEnum<::mlir::gpu::ShuffleModeAttr>($_self.getValue())";
-  let constBuilderCall = "$_builder.getStringAttr(stringifyEnum($0))";
 }
+def GPU_ShuffleModeAttr : EnumAttr<GPU_Dialect, GPU_ShuffleMode,
+                                   "shuffle_mode">;
 
+def I32OrF32 : TypeConstraint<Or<[I32.predicate, F32.predicate]>,
+                                 "i32 or f32">;
 
-def GPU_ShuffleOp : GPU_Op<"shuffle", [NoSideEffect]>,
-    Arguments<(ins AnyType:$value, I32:$offset, I32:$width,
+def GPU_ShuffleOp : GPU_Op<
+    "shuffle", [NoSideEffect, AllTypesMatch<["value", "result"]>]>,
+    Arguments<(ins I32OrF32:$value, I32:$offset, I32:$width,
                GPU_ShuffleModeAttr:$mode)>,
-    Results<(outs AnyType:$result, I1:$valid)> {
+    Results<(outs I32OrF32:$result, I1:$valid)> {
   let summary = "Shuffles values within a subgroup.";
   let description = [{
     The "shuffle" op moves values to a different invocation within the same
     subgroup.
 
     Example:
 
     ```mlir
     %1, %2 = gpu.shuffle %0, %offset, %width xor : f32
     ```
 
     For lane k returns the value from lane `k ^ offset` and `true` if that lane
     is smaller than %width. Otherwise it returns an unspecified value and
     `false`. A lane is the index of an invocation relative to its subgroup.
 
     The width specifies the number of invocations that participate in the
     shuffle. The width needs to be the same for all invocations that participate
     in the shuffle. Exactly the first `width` invocations of a subgroup need to
     execute this op in convergence.
   }];
-  let verifier = [{ return ::verifyShuffleOp(*this); }];
-  let printer = [{ printShuffleOp(p, *this); }];
-  let parser = [{ return parseShuffleOp(parser, result); }];
+  let assemblyFormat = "$mode $value `,` $offset `,` $width attr-dict `:` type($value)";
 }
 
 def GPU_BarrierOp : GPU_Op<"barrier"> {
   let summary = "Synchronizes all work items of a workgroup.";
   let description = [{
     The "barrier" op synchronizes all work items of a workgroup. It is used
     to coordinate communication between the work items of the workgroup.
 
let extraClassDeclaration = [{
     }
   }];
 
   let assemblyFormat = [{
     $value attr-dict `:` type($res)
   }];
 }
 
-def GPU_ELEMENTWISE_OP_ADD : StrEnumAttrCase<"ADDF">;
-def GPU_ELEMENTWISE_OP_MUL : StrEnumAttrCase<"MULF">;
-def GPU_ELEMENTWISE_OP_MAXF : StrEnumAttrCase<"MAXF">;
-def GPU_ELEMENTWISE_OP_MINF : StrEnumAttrCase<"MINF">;
-def GPU_ELEMENTWISE_OP_DIVF : StrEnumAttrCase<"DIVF">;
-
-def MMAElementWiseAttr : StrEnumAttr<"MMAElementwiseOp",
-  "elementwise operation to apply to mma matrix",
-  [GPU_ELEMENTWISE_OP_ADD, GPU_ELEMENTWISE_OP_MUL,
-   GPU_ELEMENTWISE_OP_MAXF, GPU_ELEMENTWISE_OP_MINF, GPU_ELEMENTWISE_OP_DIVF]> {
+def GPU_ElementwiseOpAdd  : I32EnumAttrCase<"ADDF", 0, "addf">;
+def GPU_ElementwiseOpMul  : I32EnumAttrCase<"MULF", 1, "mulf">;
+def GPU_ElementwiseOpMaxF : I32EnumAttrCase<"MAXF", 2, "maxf">;
+def GPU_ElementwiseOpMinF : I32EnumAttrCase<"MINF", 3, "minf">;
+def GPU_ElementwiseOpDivF : I32EnumAttrCase<"DIVF", 4, "divf">;
+
+def MMAElementWise : I32EnumAttr<"MMAElementwiseOp",
+  "elementwise operation to apply to mma matrix", [
+    GPU_ElementwiseOpAdd,
+    GPU_ElementwiseOpMul,
+    GPU_ElementwiseOpMaxF,
+    GPU_ElementwiseOpMinF,
+    GPU_ElementwiseOpDivF
+  ]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::gpu";
-  let storageType = "::mlir::StringAttr";
-  let returnType = "::mlir::gpu::MMAElementwiseOp";
-  let convertFromStorage = "*symbolizeMMAElementwiseOp($_self.getValue())";
-  let constBuilderCall = "$_builder.getStringAttr(stringifyEnum($0))";
 }
+def MMAElementWiseAttr : EnumAttr<GPU_Dialect, MMAElementWise,
+                                  "mma_element_wise">;
 
 def GPU_SubgroupMmaElementwiseOp : GPU_Op<"subgroup_mma_elementwise",
     [NoSideEffect,
      AllTypesMatch<["args"]>]>{
 
   let summary = "GPU warp elementwise operation on a matrix";
 
   let description = [{
let description = [{
 
     ```mlir
      %0 =  %A, %B { operation = "ADD" } :
       (!gpu.mma_matrix<16x16xf16, "COp">, !gpu.mma_matrix<16x16xf16, "COp">)
       -> !gpu.mma_matrix<16x16xf16, "COp">
     ```
   }];
 
-  let arguments = (ins Variadic<GPU_MMAMatrix>:$args, MMAElementWiseAttr:$operation);
+  let arguments = (ins Variadic<GPU_MMAMatrix>:$args,
+                       MMAElementWiseAttr:$operation);
 
   let results = (outs GPU_MMAMatrix:$res);
 
   let extraClassDeclaration = [{
     gpu::MMAMatrixType getType() {
       return res().getType().cast<gpu::MMAMatrixType>();
     }
   }];
 
   let assemblyFormat = [{
-    $args attr-dict `:` functional-type($args, $res)
+    $operation $args attr-dict `:` functional-type($args, $res)
   }];
 }
 
 #endif // GPU_OPS

mlir/include/mlir/Dialect/LLVMIR/CMakeLists.txt

 add_public_tablegen_target(MLIRLLVMConversionsIncGen)
 
 add_mlir_dialect(NVVMOps nvvm)
 add_mlir_doc(NVVMOps NVVMDialect Dialects/ -gen-dialect-doc)
 set(LLVM_TARGET_DEFINITIONS NVVMOps.td)
 mlir_tablegen(NVVMConversions.inc -gen-llvmir-conversions)
 mlir_tablegen(NVVMOpsEnums.h.inc -gen-enum-decls)
 mlir_tablegen(NVVMOpsEnums.cpp.inc -gen-enum-defs)
+mlir_tablegen(NVVMOpsAttributes.h.inc -gen-attrdef-decls -attrdefs-dialect=nvvm)
+mlir_tablegen(NVVMOpsAttributes.cpp.inc -gen-attrdef-defs -attrdefs-dialect=nvvm)
 add_public_tablegen_target(MLIRNVVMConversionsIncGen)
 
 add_mlir_dialect(ROCDLOps rocdl)
 add_mlir_doc(ROCDLOps ROCDLDialect Dialects/ -gen-dialect-doc)
 set(LLVM_TARGET_DEFINITIONS ROCDLOps.td)
 mlir_tablegen(ROCDLConversions.inc -gen-llvmir-conversions)
 add_public_tablegen_target(MLIRROCDLConversionsIncGen)
 

mlir/include/mlir/Dialect/LLVMIR/NVVMDialect.h

 /// Return the element type and number of elements associated with a wmma matrix
 /// of given chracteristics. This matches the logic in IntrinsicsNVVM.td
 /// WMMA_REGS structure.
 std::pair<mlir::Type, unsigned> inferMMAType(mlir::NVVM::MMATypes type,
                                              mlir::NVVM::MMAFrag frag,
                                              mlir::MLIRContext *context);
 
 ///// Ops /////
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/LLVMIR/NVVMOpsAttributes.h.inc"
+
 #define GET_OP_CLASSES
 #include "mlir/Dialect/LLVMIR/NVVMOps.h.inc"
 
 #include "mlir/Dialect/LLVMIR/NVVMOpsDialect.h.inc"
 
 #endif /* MLIR_DIALECT_LLVMIR_NVVMDIALECT_H_ */

mlir/include/mlir/Dialect/LLVMIR/NVVMOps.td

 //===-- NVVMOps.td - NVVM IR dialect op definition file ----*- tablegen -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This is the NVVM IR operation definition file.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef NVVMIR_OPS
 #define NVVMIR_OPS
 
+include "mlir/IR/EnumAttr.td"
 include "mlir/Dialect/LLVMIR/LLVMOpBase.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
 
 def LLVM_i8Ptr_global : LLVM_IntPtrBase<8, 1>;
 def LLVM_i8Ptr_shared : LLVM_IntPtrBase<8, 3>;
 
 //===----------------------------------------------------------------------===//
 // NVVM dialect definitions
 //===----------------------------------------------------------------------===//
 
 def NVVM_Dialect : Dialect {
   let name = "nvvm";
   let cppNamespace = "::mlir::NVVM";
   let dependentDialects = ["LLVM::LLVMDialect"];
   let hasOperationAttrVerify = 1;
 
   let extraClassDeclaration = [{
     /// Get the name of the attribute used to annotate external kernel
     /// functions.
     static StringRef getKernelFuncAttrName() { return "nvvm.kernel"; }
   }];
+
+  let useDefaultAttributePrinterParser = 1;
 }
 
 //===----------------------------------------------------------------------===//
 // NVVM op definitions
 //===----------------------------------------------------------------------===//
 
 class NVVM_Op<string mnemonic, list<OpTrait> traits = []> :
   LLVM_OpBase<NVVM_Dialect, mnemonic, traits> {

 
 def NVVM_Barrier0Op : NVVM_Op<"barrier0"> {
   string llvmBuilder = [{
       createIntrinsicCall(builder, llvm::Intrinsic::nvvm_barrier0);
   }];
   let assemblyFormat = "attr-dict";
 }
 
-def ShflKindBfly : StrEnumAttrCase<"bfly">;
-def ShflKindUp : StrEnumAttrCase<"up">;
-def ShflKindDown : StrEnumAttrCase<"down">;
-def ShflKindIdx : StrEnumAttrCase<"idx">;
+def ShflKindBfly : I32EnumAttrCase<"bfly", 0>;
+def ShflKindUp   : I32EnumAttrCase<"up", 1>;
+def ShflKindDown : I32EnumAttrCase<"down", 2>;
+def ShflKindIdx  : I32EnumAttrCase<"idx", 3>;
 
 /// Enum attribute of the different shuffle kinds.
-def ShflKind : StrEnumAttr<"ShflKind", "NVVM shuffle kind",
+def ShflKind : I32EnumAttr<"ShflKind", "NVVM shuffle kind",
   [ShflKindBfly, ShflKindUp, ShflKindDown, ShflKindIdx]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::NVVM";
-  let storageType = "mlir::StringAttr";
-  let returnType = "NVVM::ShflKind";
-  let convertFromStorage = "*symbolizeEnum<NVVM::ShflKind>($_self.getValue())";
-  let constBuilderCall = "$_builder.getStringAttr(stringifyEnum($0))";
 }
+def ShflKindAttr : EnumAttr<NVVM_Dialect, ShflKind, "shfl_kind">;
 
 def NVVM_ShflOp :
   NVVM_Op<"shfl.sync">,
   Results<(outs LLVM_Type:$res)>,
   Arguments<(ins I32:$dst,
                  LLVM_Type:$val,
                  I32:$offset,
                  I32:$mask_and_clamp,
-                 ShflKind:$kind,
+                 ShflKindAttr:$kind,
                  OptionalAttr<UnitAttr>:$return_value_and_is_valid)> {
   string llvmBuilder = [{
       auto intId = getShflIntrinsicId(
           $_resultType, $kind, static_cast<bool>($return_value_and_is_valid));
       $res = createIntrinsicCall(builder,
           intId, {$dst, $val, $offset, $mask_and_clamp});
   }];
   let verifier = [{
class MMA_MMA_INTR<string opName> {
   "  return " #
        WMMA_NAME<opName, layoutA, layoutB, op[0], op[1], op[2], op[3]>.id # ";")));
   list<string> f = !foldl([""],
                      !foldl([[""]], cond0, acc, el, !listconcat(acc, el)),
                           acc1, el1, !listconcat(acc1, el1));
   string id = !foldl("", f, acc, el, acc # "\n" # el);
 }
 
-def MMALayoutRow : StrEnumAttrCase<"row">;
-def MMALayoutCol : StrEnumAttrCase<"col">;
+def MMALayoutRow : I32EnumAttrCase<"row", 0>;
+def MMALayoutCol : I32EnumAttrCase<"col", 1>;
 
 /// Enum attribute of the different matrix layout.
-def MMALayout : StrEnumAttr<"MMALayout", "NVVM MMA layout",
+def MMALayout : I32EnumAttr<"MMALayout", "NVVM MMA layout",
   [MMALayoutRow, MMALayoutCol]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::NVVM";
-  let storageType = "mlir::StringAttr";
-  let returnType = "NVVM::MMALayout";
-  let convertFromStorage = "*symbolizeEnum<NVVM::MMALayout>($_self.getValue())";
-  let constBuilderCall = "$_builder.getStringAttr(stringifyEnum($0))";
+}
+def MMALayoutAttr : EnumAttr<NVVM_Dialect, MMALayout, "mma_layout"> {
+  let assemblyFormat = "`<` $value `>`";
 }
 
-def MMATypeF16 : StrEnumAttrCase<"f16">;
-def MMATypeF32 : StrEnumAttrCase<"f32">;
-def MMATypeTF32 : StrEnumAttrCase<"tf32">;
+def MMATypeF16  : I32EnumAttrCase<"f16", 0>;
+def MMATypeF32  : I32EnumAttrCase<"f32", 1>;
+def MMATypeTF32 : I32EnumAttrCase<"tf32", 2>;
 
 /// Enum attribute of the different matrix types.
-def MMATypes : StrEnumAttr<"MMATypes", "NVVM MMA types",
+def MMATypes : I32EnumAttr<"MMATypes", "NVVM MMA types",
   [MMATypeF16, MMATypeF32, MMATypeTF32]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::NVVM";
-  let storageType = "mlir::StringAttr";
-  let returnType = "NVVM::MMATypes";
-  let convertFromStorage = "*symbolizeEnum<NVVM::MMATypes>($_self.getValue())";
-  let constBuilderCall = "$_builder.getStringAttr(stringifyEnum($0))";
+}
+def MMATypesAttr : EnumAttr<NVVM_Dialect, MMATypes, "mma_type"> {
+  let assemblyFormat = "`<` $value `>`";
 }
 
-def MMAFragA : StrEnumAttrCase<"a">;
-def MMAFragB : StrEnumAttrCase<"b">;
-def MMAFragC : StrEnumAttrCase<"c">;
+def MMAFragA : I32EnumAttrCase<"a", 0>;
+def MMAFragB : I32EnumAttrCase<"b", 1>;
+def MMAFragC : I32EnumAttrCase<"c", 2>;
 
 /// Enum attribute of the different frag types.
-def MMAFragAttr : StrEnumAttr<"MMAFrag", "NVVM MMA frag type",
+def MMAFrag: I32EnumAttr<"MMAFrag", "NVVM MMA frag type",
   [MMAFragA, MMAFragB, MMAFragC]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::NVVM";
-  let storageType = "mlir::StringAttr";
-  let returnType = "NVVM::MMAFrag";
-  let convertFromStorage = "*symbolizeEnum<NVVM::MMAFrag>($_self.getValue())";
-  let constBuilderCall = "$_builder.getStringAttr(stringifyEnum($0))";
+}
+def MMAFragAttr : EnumAttr<NVVM_Dialect, MMAFrag, "mma_frag"> {
+  let assemblyFormat = "`<` $value `>`";
 }
 
 def NVVM_WMMALoadOp: NVVM_Op<"wmma.load">,
   Results<(outs LLVM_AnyStruct:$res)>,
   Arguments<(ins LLVM_AnyPointer: $ptr, I32: $stride, I32Attr:$m,
-             I32Attr:$n, I32Attr:$k, MMALayout:$layout, MMATypes:$eltype,
-             MMAFragAttr:$frag)> {
+             I32Attr:$n, I32Attr:$k, MMALayoutAttr:$layout,
+             MMATypesAttr:$eltype, MMAFragAttr:$frag)> {
 
   let summary = "Warp synchronous matrix load";
 
   // Since LLVM intrinsic IDs are enum that cannot be dynamically generated in
   // C++ we instanciate a function in tablegen to map the valide configuration
   // to the corresponsding intrinsic ID.
   // Because we want a single source of truth, this mean the source of truth
   // about valid combinations needs to be in tablgen, therefore we generate
string baseDescription = [{
     }];
 
   let assemblyFormat = "$ptr `,` $stride attr-dict `:` functional-type($ptr, $res)";
   let verifier = [{ return ::verify(*this); }];
 }
 
 def NVVM_WMMAStoreOp : NVVM_Op<"wmma.store">,
   Arguments<(ins LLVM_AnyPointer: $ptr,
-             I32Attr:$m, I32Attr:$n, I32Attr:$k, MMALayout:$layout,
-             MMATypes:$eltype, Variadic<LLVM_Type>:$args, I32: $stride)>{
+             I32Attr:$m, I32Attr:$n, I32Attr:$k, MMALayoutAttr:$layout,
+             MMATypesAttr:$eltype, Variadic<LLVM_Type>:$args, I32: $stride)>{
   let summary = "Warp synchronous matrix store";
 
   let extraClassDeclaration =
     "static llvm::Intrinsic::ID getIntrinsicID("
     "int m, int n, int k, mlir::NVVM::MMALayout layoutEnum,"
     "mlir::NVVM::MMATypes eltypeEnum) {"
     "  llvm::StringRef layout = stringifyEnum(layoutEnum);"
     "  llvm::StringRef eltype = stringifyEnum(eltypeEnum);"
def NVVM_WMMAStoreOp : NVVM_Op<"wmma.store">,
 
   let assemblyFormat = "$ptr `,` $stride `,` $args attr-dict `:` type($ptr) `,` type($args)";
   let verifier = [{ return ::verify(*this); }];
 }
 
 // Base class for all the variants of WMMA mmaOps that may be defined.
 def NVVM_WMMAMmaOp : NVVM_Op<"wmma.mma">,
   Results<(outs LLVM_AnyStruct:$res)>,
-  Arguments<(ins I32Attr:$m, I32Attr:$n, I32Attr:$k, MMALayout:$layoutA,
-            MMALayout:$layoutB, MMATypes:$eltypeA, MMATypes:$eltypeB,
-            Variadic<LLVM_Type>:$args)>{
+  Arguments<(ins I32Attr:$m, I32Attr:$n, I32Attr:$k, MMALayoutAttr:$layoutA,
+             MMALayoutAttr:$layoutB, MMATypesAttr:$eltypeA,
+             MMATypesAttr:$eltypeB, Variadic<LLVM_Type>:$args)>{
   let summary = "Warp synchronous matrix-multiply accumulate using tensor cores.";
 
   let extraClassDeclaration =
     "static llvm::Intrinsic::ID getIntrinsicID("
     "int m, int n, int k, mlir::NVVM::MMALayout layoutAEnum,"
     "mlir::NVVM::MMALayout layoutBEnum, mlir::NVVM::MMATypes eltypeAEnum,"
     "mlir::NVVM::MMATypes eltypeBEnum) {"
     "llvm::StringRef layoutA = stringifyEnum(layoutAEnum);"

mlir/include/mlir/Dialect/OpenACC/CMakeLists.txt

 set(LLVM_TARGET_DEFINITIONS ${LLVM_MAIN_INCLUDE_DIR}/llvm/Frontend/OpenACC/ACC.td)
 mlir_tablegen(AccCommon.td --gen-directive-decl)
 add_public_tablegen_target(acc_common_td)
 
 set(LLVM_TARGET_DEFINITIONS OpenACCOps.td)
 mlir_tablegen(OpenACCOpsDialect.h.inc -gen-dialect-decls -dialect=acc)
 mlir_tablegen(OpenACCOpsDialect.cpp.inc -gen-dialect-defs -dialect=acc)
 mlir_tablegen(OpenACCOps.h.inc -gen-op-decls)
 mlir_tablegen(OpenACCOps.cpp.inc -gen-op-defs)
 mlir_tablegen(OpenACCOpsEnums.h.inc -gen-enum-decls)
 mlir_tablegen(OpenACCOpsEnums.cpp.inc -gen-enum-defs)
+mlir_tablegen(OpenACCOpsAttributes.h.inc -gen-attrdef-decls -attrdefs-dialect=acc)
+mlir_tablegen(OpenACCOpsAttributes.cpp.inc -gen-attrdef-defs -attrdefs-dialect=acc)
 add_mlir_doc(OpenACCOps OpenACCDialect Dialects/ -gen-dialect-doc)
 add_public_tablegen_target(MLIROpenACCOpsIncGen)
 add_dependencies(OpenACCDialectDocGen acc_common_td)

mlir/include/mlir/Dialect/OpenACC/OpenACC.h

 #define MLIR_DIALECT_OPENACC_OPENACC_H_
 
 #include "mlir/IR/Dialect.h"
 #include "mlir/IR/OpDefinition.h"
 
 #include "mlir/Dialect/OpenACC/OpenACCOpsDialect.h.inc"
 #include "mlir/Dialect/OpenACC/OpenACCOpsEnums.h.inc"
 
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/OpenACC/OpenACCOpsAttributes.h.inc"
+
 #define GET_OP_CLASSES
 #include "mlir/Dialect/OpenACC/OpenACCOps.h.inc"
 
 namespace mlir {
 namespace acc {
 
 /// Enumeration used to encode the execution mapping on a loop construct.
 /// They refer directly to the OpenACC 3.1 standard:

mlir/include/mlir/Dialect/OpenACC/OpenACCOps.td

 //===- OpenACC.td - OpenACC operation definitions ----------*- tablegen -*-===//
 //
 // Part of the MLIR Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 // =============================================================================
 //
 // Defines MLIR OpenACC operations.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef OPENACC_OPS
 #define OPENACC_OPS
 
+include "mlir/IR/EnumAttr.td"
 include "mlir/IR/OpBase.td"
 include "mlir/Dialect/OpenACC/AccCommon.td"
 
 def OpenACC_Dialect : Dialect {
   let name = "acc";
 
   let summary = "An OpenACC dialect for MLIR.";
 
   let description = [{
     This dialect models the construct from the OpenACC 3.1 directive language.
   }];
 
+  let useDefaultAttributePrinterParser = 1;
   let cppNamespace = "::mlir::acc";
 }
 
 // Base class for OpenACC dialect ops.
 class OpenACC_Op<string mnemonic, list<OpTrait> traits = []> :
   Op<OpenACC_Dialect, mnemonic, traits> {
 
   let printer = [{ return ::print(p, *this); }];
   let verifier = [{ return ::verify(*this); }];
   let parser = [{ return ::parse$cppClass(parser, result); }];
 }
 
 // Reduction operation enumeration.
-def OpenACC_ReductionOpAdd : StrEnumAttrCase<"redop_add">;
-def OpenACC_ReductionOpMul : StrEnumAttrCase<"redop_mul">;
-def OpenACC_ReductionOpMax : StrEnumAttrCase<"redop_max">;
-def OpenACC_ReductionOpMin : StrEnumAttrCase<"redop_min">;
-def OpenACC_ReductionOpAnd : StrEnumAttrCase<"redop_and">;
-def OpenACC_ReductionOpOr : StrEnumAttrCase<"redop_or">;
-def OpenACC_ReductionOpXor : StrEnumAttrCase<"redop_xor">;
-def OpenACC_ReductionOpLogEqv : StrEnumAttrCase<"redop_leqv">;
-def OpenACC_ReductionOpLogNeqv : StrEnumAttrCase<"redop_lneqv">;
-def OpenACC_ReductionOpLogAnd : StrEnumAttrCase<"redop_land">;
-def OpenACC_ReductionOpLogOr : StrEnumAttrCase<"redop_lor">;
+def OpenACC_ReductionOpAdd     : I32EnumAttrCase<"redop_add", 0>;
+def OpenACC_ReductionOpMul     : I32EnumAttrCase<"redop_mul", 1>;
+def OpenACC_ReductionOpMax     : I32EnumAttrCase<"redop_max", 2>;
+def OpenACC_ReductionOpMin     : I32EnumAttrCase<"redop_min", 3>;
+def OpenACC_ReductionOpAnd     : I32EnumAttrCase<"redop_and", 4>;
+def OpenACC_ReductionOpOr      : I32EnumAttrCase<"redop_or", 5>;
+def OpenACC_ReductionOpXor     : I32EnumAttrCase<"redop_xor", 6>;
+def OpenACC_ReductionOpLogEqv  : I32EnumAttrCase<"redop_leqv", 7>;
+def OpenACC_ReductionOpLogNeqv : I32EnumAttrCase<"redop_lneqv", 8>;
+def OpenACC_ReductionOpLogAnd  : I32EnumAttrCase<"redop_land", 9>;
+def OpenACC_ReductionOpLogOr   : I32EnumAttrCase<"redop_lor", 10>;
 
-def OpenACC_ReductionOpAttr : StrEnumAttr<"ReductionOpAttr",
+def OpenACC_ReductionOp : I32EnumAttr<"ReductionOp",
     "built-in reduction operations supported by OpenACC",
     [OpenACC_ReductionOpAdd, OpenACC_ReductionOpMul, OpenACC_ReductionOpMax,
      OpenACC_ReductionOpMin, OpenACC_ReductionOpAnd, OpenACC_ReductionOpOr,
      OpenACC_ReductionOpXor, OpenACC_ReductionOpLogEqv,
      OpenACC_ReductionOpLogNeqv, OpenACC_ReductionOpLogAnd,
      OpenACC_ReductionOpLogOr
     ]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::acc";
 }
+def OpenACC_ReductionOpAttr : EnumAttr<OpenACC_Dialect, OpenACC_ReductionOp,
+                                       "reduction_op">;
 
 // Type used in operation below.
 def IntOrIndex : AnyTypeOf<[AnyInteger, Index]>;
 
 //===----------------------------------------------------------------------===//
 // 2.5.1 parallel Construct
 //===----------------------------------------------------------------------===//
 

mlir/include/mlir/Dialect/OpenMP/CMakeLists.txt

 set(LLVM_TARGET_DEFINITIONS ${LLVM_MAIN_INCLUDE_DIR}/llvm/Frontend/OpenMP/OMP.td)
 mlir_tablegen(OmpCommon.td --gen-directive-decl)
 add_public_tablegen_target(omp_common_td)
 
 set(LLVM_TARGET_DEFINITIONS OpenMPOps.td)
 mlir_tablegen(OpenMPOpsDialect.h.inc -gen-dialect-decls -dialect=omp)
 mlir_tablegen(OpenMPOpsDialect.cpp.inc -gen-dialect-defs -dialect=omp)
 mlir_tablegen(OpenMPOps.h.inc -gen-op-decls)
 mlir_tablegen(OpenMPOps.cpp.inc -gen-op-defs)
 mlir_tablegen(OpenMPOpsEnums.h.inc -gen-enum-decls)
 mlir_tablegen(OpenMPOpsEnums.cpp.inc -gen-enum-defs)
 mlir_tablegen(OpenMPTypeInterfaces.h.inc -gen-type-interface-decls)
 mlir_tablegen(OpenMPTypeInterfaces.cpp.inc -gen-type-interface-defs)
+mlir_tablegen(OpenMPOpsAttributes.h.inc -gen-attrdef-decls -attrdefs-dialect=omp)
+mlir_tablegen(OpenMPOpsAttributes.cpp.inc -gen-attrdef-defs -attrdefs-dialect=omp)
 add_mlir_doc(OpenMPOps OpenMPDialect Dialects/ -gen-dialect-doc)
 add_public_tablegen_target(MLIROpenMPOpsIncGen)
 add_dependencies(OpenMPDialectDocGen omp_common_td)
 add_mlir_interface(OpenMPOpsInterfaces)

mlir/include/mlir/Dialect/OpenMP/OpenMPDialect.h

 #include "mlir/Interfaces/ControlFlowInterfaces.h"
 #include "mlir/Interfaces/SideEffectInterfaces.h"
 
 #include "mlir/Dialect/OpenMP/OpenMPOpsDialect.h.inc"
 #include "mlir/Dialect/OpenMP/OpenMPOpsEnums.h.inc"
 #include "mlir/Dialect/OpenMP/OpenMPOpsInterfaces.h.inc"
 #include "mlir/Dialect/OpenMP/OpenMPTypeInterfaces.h.inc"
 
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/OpenMP/OpenMPOpsAttributes.h.inc"
+
 #define GET_OP_CLASSES
 #include "mlir/Dialect/OpenMP/OpenMPOps.h.inc"
 
 #endif // MLIR_DIALECT_OPENMP_OPENMPDIALECT_H_

mlir/include/mlir/Dialect/OpenMP/OpenMPOps.td

 //===-- OpenMPOps.td - OpenMP dialect operation definitions *- tablegen -*-===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file defines the basic operations for the OpenMP dialect.
 //
 //===----------------------------------------------------------------------===//
 
 
 #ifndef OPENMP_OPS
 #define OPENMP_OPS
 
+include "mlir/IR/EnumAttr.td"
 include "mlir/IR/OpBase.td"
 include "mlir/Interfaces/SideEffectInterfaces.td"
 include "mlir/Interfaces/ControlFlowInterfaces.td"
 include "mlir/IR/SymbolInterfaces.td"
 include "mlir/Dialect/OpenMP/OmpCommon.td"
 include "mlir/Dialect/LLVMIR/LLVMOpBase.td"
 include "mlir/Dialect/OpenMP/OpenMPOpsInterfaces.td"
 
 def OpenMP_Dialect : Dialect {
   let name = "omp";
   let cppNamespace = "::mlir::omp";
   let dependentDialects = ["::mlir::LLVM::LLVMDialect"];
+  let useDefaultAttributePrinterParser = 1;
 }
 
 class OpenMP_Op<string mnemonic, list<OpTrait> traits = []> :
       Op<OpenMP_Dialect, mnemonic, traits>;
 
 // Type which can be constraint accepting standard integers and indices.
 def IntLikeType : AnyTypeOf<[AnyInteger, Index]>;
 
def OpenMP_PointerLikeType : Type<
   CPred<"$_self.isa<::mlir::omp::PointerLikeType>()">,
   "OpenMP-compatible variable type", "::mlir::omp::PointerLikeType">;
 
 //===----------------------------------------------------------------------===//
 // 2.6 parallel Construct
 //===----------------------------------------------------------------------===//
 
 // Possible values for the default clause
-def ClauseDefaultPrivate : StrEnumAttrCase<"defprivate">;
-def ClauseDefaultFirstPrivate : StrEnumAttrCase<"deffirstprivate">;
-def ClauseDefaultShared : StrEnumAttrCase<"defshared">;
-def ClauseDefaultNone : StrEnumAttrCase<"defnone">;
+def ClauseDefaultPrivate      : I32EnumAttrCase<"defprivate",      0>;
+def ClauseDefaultFirstPrivate : I32EnumAttrCase<"deffirstprivate", 1>;
+def ClauseDefaultShared       : I32EnumAttrCase<"defshared",       2>;
+def ClauseDefaultNone         : I32EnumAttrCase<"defnone",         3>;
 
-def ClauseDefault : StrEnumAttr<
+def ClauseDefault : I32EnumAttr<
     "ClauseDefault",
     "default clause",
     [ClauseDefaultPrivate, ClauseDefaultFirstPrivate, ClauseDefaultShared,
      ClauseDefaultNone]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::omp";
 }
+def ClauseDefaultAttr : EnumAttr<OpenMP_Dialect, ClauseDefault,
+                                 "clause_default">;
 
 def ParallelOp : OpenMP_Op<"parallel", [AttrSizedOperandSegments,
                  DeclareOpInterfaceMethods<OutlineableOpenMPOpInterface>]> {
   let summary = "parallel construct";
   let description = [{
     The parallel construct includes a region of code which is to be executed
     by a team of threads.
 
let description = [{
     that specify the memory allocator to be used to obtain storage for private values.
 
     The optional $proc_bind_val attribute controls the thread affinity for the execution
     of the parallel region.
   }];
 
   let arguments = (ins Optional<AnyType>:$if_expr_var,
              Optional<AnyType>:$num_threads_var,
-             OptionalAttr<ClauseDefault>:$default_val,
+             OptionalAttr<ClauseDefaultAttr>:$default_val,
              Variadic<AnyType>:$private_vars,
              Variadic<AnyType>:$firstprivate_vars,
              Variadic<AnyType>:$shared_vars,
              Variadic<AnyType>:$copyin_vars,
              Variadic<AnyType>:$allocate_vars,
              Variadic<AnyType>:$allocators_vars,
              OptionalAttr<ProcBindKind>:$proc_bind_val);
 
let description = [{
     operation.  These regions are not expected to return any value so the
     terminator takes no operands. The terminator op returns control to the
     enclosing op.
   }];
 
   let assemblyFormat = "attr-dict";
 }
 
-def OMP_SCHEDULE_MOD_None : StrEnumAttrCase<"none", 0>;
-def OMP_SCHEDULE_MOD_Monotonic : StrEnumAttrCase<"monotonic", 1>;
-def OMP_SCHEDULE_MOD_Nonmonotonic : StrEnumAttrCase<"nonmonotonic", 2>;
-def OMP_SCHEDULE_MOD_SIMD : StrEnumAttrCase<"simd", 3>;
-
-def ScheduleModifier : StrEnumAttr<"ScheduleModifier", "OpenMP Schedule Modifier",
-                                         [OMP_SCHEDULE_MOD_None,
-                                          OMP_SCHEDULE_MOD_Monotonic,
-                                          OMP_SCHEDULE_MOD_Nonmonotonic,
-                                          OMP_SCHEDULE_MOD_SIMD]>
-{
-    let cppNamespace = "::mlir::omp";
+def OMP_ScheduleModNone         : I32EnumAttrCase<"none", 0>;
+def OMP_ScheduleModMonotonic    : I32EnumAttrCase<"monotonic", 1>;
+def OMP_ScheduleModNonmonotonic : I32EnumAttrCase<"nonmonotonic", 2>;
+def OMP_ScheduleModSIMD         : I32EnumAttrCase<"simd", 3>;
+
+def ScheduleModifier
+    : I32EnumAttr<"ScheduleModifier", "OpenMP Schedule Modifier",
+                  [OMP_ScheduleModNone, OMP_ScheduleModMonotonic,
+                   OMP_ScheduleModNonmonotonic, OMP_ScheduleModSIMD]> {
+  let genSpecializedAttr = 0;
 }
+def ScheduleModifierAttr : EnumAttr<OpenMP_Dialect, ScheduleModifier,
+                                    "sched_mod">;
 
 //===----------------------------------------------------------------------===//
 // 2.8.1 Sections Construct
 //===----------------------------------------------------------------------===//
 
 def SectionOp : OpenMP_Op<"section", [HasParent<"SectionsOp">]> {
   let summary = "section directive";
   let description = [{
let arguments = (ins Variadic<IntLikeType>:$lowerBound,
              Variadic<AnyType>:$firstprivate_vars,
              Variadic<AnyType>:$lastprivate_vars,
              Variadic<AnyType>:$linear_vars,
              Variadic<AnyType>:$linear_step_vars,
              Variadic<OpenMP_PointerLikeType>:$reduction_vars,
              OptionalAttr<SymbolRefArrayAttr>:$reductions,
              OptionalAttr<ScheduleKind>:$schedule_val,
              Optional<AnyType>:$schedule_chunk_var,
-             OptionalAttr<ScheduleModifier>:$schedule_modifier,
+             OptionalAttr<ScheduleModifierAttr>:$schedule_modifier,
              UnitAttr:$simd_modifier,
              Confined<OptionalAttr<I64Attr>, [IntMinValue<0>]>:$collapse_val,
              UnitAttr:$nowait,
              Confined<OptionalAttr<I64Attr>, [IntMinValue<0>]>:$ordered_val,
              OptionalAttr<OrderKind>:$order_val,
              UnitAttr:$inclusive);
 
   let skipDefaultBuilders = 1;
def BarrierOp : OpenMP_Op<"barrier"> {
 
   let assemblyFormat = "attr-dict";
 }
 
 //===----------------------------------------------------------------------===//
 // [5.1] 2.19.9 ordered Construct
 //===----------------------------------------------------------------------===//
 
-def ClauseDependSource : StrEnumAttrCase<"dependsource">;
-def ClauseDependSink : StrEnumAttrCase<"dependsink">;
+def ClauseDependSource : I32EnumAttrCase<"dependsource", 0>;
+def ClauseDependSink   : I32EnumAttrCase<"dependsink",   1>;
 
-def ClauseDepend : StrEnumAttr<
+def ClauseDepend : I32EnumAttr<
     "ClauseDepend",
     "depend clause",
     [ClauseDependSource, ClauseDependSink]> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::omp";
 }
+def ClauseDependAttr : EnumAttr<OpenMP_Dialect, ClauseDepend, "clause_depend"> {
+  let assemblyFormat = "`(` $value `)`";
+}
 
 def OrderedOp : OpenMP_Op<"ordered"> {
   let summary = "ordered construct without region";
   let description = [{
     The ordered construct without region is a stand-alone directive that
     specifies cross-iteration dependences in a doacross loop nest.
 
     The `depend_type_val` attribute refers to either the DEPEND(SOURCE) clause
     or the DEPEND(SINK: vec) clause.
 
     The `num_loops_val` attribute specifies the number of loops in the doacross
     nest.
 
     The `depend_vec_vars` is a variadic list of operands that specifies the index
     of the loop iterator in the doacross nest for the DEPEND(SOURCE) clause or
     the index of the element of "vec" for the DEPEND(SINK: vec) clause. It
     contains the operands in multiple "vec" when multiple DEPEND(SINK: vec)
     clauses exist in one ORDERED directive.
   }];
 
-  let arguments = (ins OptionalAttr<ClauseDepend>:$depend_type_val,
+  let arguments = (ins OptionalAttr<ClauseDependAttr>:$depend_type_val,
              Confined<OptionalAttr<I64Attr>, [IntMinValue<0>]>:$num_loops_val,
              Variadic<AnyType>:$depend_vec_vars);
 
   let assemblyFormat = [{
-    ( `depend_type` `(` $depend_type_val^ `)` )?
+    ( `depend_type` `` $depend_type_val^ )?
     ( `depend_vec` `(` $depend_vec_vars^ `:` type($depend_vec_vars) `)` )?
     attr-dict
   }];
 
   let verifier = "return ::verifyOrderedOp(*this);";
 }
 
 def OrderedRegionOp : OpenMP_Op<"ordered_region"> {

mlir/include/mlir/Dialect/SPIRV/IR/SPIRVAttributes.h

 #include "mlir/Support/LLVM.h"
 
 // Pull in TableGen'erated SPIR-V attribute definitions for target and ABI.
 #include "mlir/Dialect/SPIRV/IR/TargetAndABI.h.inc"
 
 namespace mlir {
 namespace spirv {
 enum class Capability : uint32_t;
-enum class DeviceType;
-enum class Extension;
-enum class Vendor;
+enum class DeviceType : uint32_t;
+enum class Extension : uint32_t;
+enum class Vendor : uint32_t;
 enum class Version : uint32_t;
 
 namespace detail {
 struct InterfaceVarABIAttributeStorage;
 struct TargetEnvAttributeStorage;
 struct VerCapExtAttributeStorage;
 } // namespace detail
 

mlir/include/mlir/Dialect/SPIRV/IR/SPIRVBase.td

 // This file defines the SPIR-V dialect, common SPIR-V types, and utilities
 // for facilitating defining SPIR-V ops.
 //
 //===----------------------------------------------------------------------===//
 
 #ifndef MLIR_DIALECT_SPIRV_IR_BASE
 #define MLIR_DIALECT_SPIRV_IR_BASE
 
+include "mlir/IR/EnumAttr.td"
 include "mlir/IR/OpBase.td"
 include "mlir/Dialect/SPIRV/IR/SPIRVAvailability.td"
 
 //===----------------------------------------------------------------------===//
 // SPIR-V dialect definitions
 //===----------------------------------------------------------------------===//
 
 def SPIRV_Dialect : Dialect {
class SPV_I32EnumAttr<string name, string description,
   let predicate = And<[
     I32Attr.predicate,
     SPV_IsKnownEnumCaseFor<name>,
   ]>;
   let cppNamespace = "::mlir::spirv";
 }
 
 // Wrapper over base StrEnumAttr to set common fields.
-class SPV_StrEnumAttr<string name, string description,
-                      list<StrEnumAttrCase> cases> :
-    StrEnumAttr<name, description, cases> {
-  let predicate = And<[
-    StrAttr.predicate,
-    CPred<"::mlir::spirv::symbolize" # name # "("
-          "$_self.cast<StringAttr>().getValue()).hasValue()">,
-  ]>;
+class SPV_Enum<string name, string description, list<I32EnumAttrCase> cases>
+    : I32EnumAttr<name, description, cases> {
+  let genSpecializedAttr = 0;
   let cppNamespace = "::mlir::spirv";
 }
+class SPV_EnumAttr<string name, string description, string mnemonic,
+                      list<I32EnumAttrCase> cases> :
+    EnumAttr<SPIRV_Dialect, SPV_Enum<name, description, cases>, mnemonic>;
 
 //===----------------------------------------------------------------------===//
 // SPIR-V availability definitions
 //===----------------------------------------------------------------------===//
 
 def SPV_V_1_0 : I32EnumAttrCase<"V_1_0", 0, "v1.0">;
 def SPV_V_1_1 : I32EnumAttrCase<"V_1_1", 1, "v1.1">;
 def SPV_V_1_2 : I32EnumAttrCase<"V_1_2", 2, "v1.2">;
let interfaceDescription = [{
     Querying interface for maximal supported SPIR-V version.
 
     This interface provides a `getMaxVersion()` method to query the maximal
     supported version for the implementing SPIR-V operation. The returned value
     is a `mlir::spirv::Version` enumerant.
   }];
 }
 
-class Extension<list<StrEnumAttrCase> extensions> : Availability {
+class Extension<list<I32EnumAttrCase> extensions> : Availability {
   let cppNamespace = "::mlir::spirv";
   let interfaceName = "QueryExtensionInterface";
   let interfaceDescription = [{
     Querying interface for required SPIR-V extensions.
 
     This interface provides a `getExtensions()` method to query the required
     extensions for the implementing SPIR-V operation. The returned value
     is a nested vector whose element is `mlir::spirv::Extension`s. The outer
let methods = [InterfaceMethod<
     "::llvm::SmallVector<::llvm::ArrayRef<::mlir::spirv::Capability>, 1>",
     "getCapabilities">];
 }
 
 //===----------------------------------------------------------------------===//
 // SPIR-V target GPU vendor and device definitions
 //===----------------------------------------------------------------------===//
 
-def SPV_DT_CPU           : StrEnumAttrCase<"CPU">;
-def SPV_DT_DiscreteGPU   : StrEnumAttrCase<"DiscreteGPU">;
-def SPV_DT_IntegratedGPU : StrEnumAttrCase<"IntegratedGPU">;
+def SPV_DT_CPU           : I32EnumAttrCase<"CPU", 0>;
+def SPV_DT_DiscreteGPU   : I32EnumAttrCase<"DiscreteGPU", 1>;
+def SPV_DT_IntegratedGPU : I32EnumAttrCase<"IntegratedGPU", 2>;
 // An accelerator other than GPU or CPU
-def SPV_DT_Other         : StrEnumAttrCase<"Other">;
+def SPV_DT_Other         : I32EnumAttrCase<"Other", 3>;
 // Information missing.
-def SPV_DT_Unknown       : StrEnumAttrCase<"Unknown">;
+def SPV_DT_Unknown       : I32EnumAttrCase<"Unknown", 4>;
 
-def SPV_DeviceTypeAttr : SPV_StrEnumAttr<
-  "DeviceType", "valid SPIR-V device types", [
+def SPV_DeviceTypeAttr : SPV_EnumAttr<
+  "DeviceType", "valid SPIR-V device types", "device_type", [
     SPV_DT_Other, SPV_DT_IntegratedGPU, SPV_DT_DiscreteGPU,
     SPV_DT_CPU, SPV_DT_Unknown
   ]>;
 
-def SPV_V_AMD         : StrEnumAttrCase<"AMD">;
-def SPV_V_ARM         : StrEnumAttrCase<"ARM">;
-def SPV_V_Imagination : StrEnumAttrCase<"Imagination">;
-def SPV_V_Intel       : StrEnumAttrCase<"Intel">;
-def SPV_V_NVIDIA      : StrEnumAttrCase<"NVIDIA">;
-def SPV_V_Qualcomm    : StrEnumAttrCase<"Qualcomm">;
-def SPV_V_SwiftShader : StrEnumAttrCase<"SwiftShader">;
-def SPV_V_Unknown     : StrEnumAttrCase<"Unknown">;
+def SPV_V_AMD         : I32EnumAttrCase<"AMD", 0>;
+def SPV_V_ARM         : I32EnumAttrCase<"ARM", 1>;
+def SPV_V_Imagination : I32EnumAttrCase<"Imagination", 2>;
+def SPV_V_Intel       : I32EnumAttrCase<"Intel", 3>;
+def SPV_V_NVIDIA      : I32EnumAttrCase<"NVIDIA", 4>;
+def SPV_V_Qualcomm    : I32EnumAttrCase<"Qualcomm", 5>;
+def SPV_V_SwiftShader : I32EnumAttrCase<"SwiftShader", 6>;
+def SPV_V_Unknown     : I32EnumAttrCase<"Unknown", 7>;
 
-def SPV_VendorAttr : SPV_StrEnumAttr<
-  "Vendor", "recognized SPIR-V vendor strings", [
+def SPV_VendorAttr : SPV_EnumAttr<
+  "Vendor", "recognized SPIR-V vendor strings", "vendor", [
     SPV_V_AMD, SPV_V_ARM, SPV_V_Imagination, SPV_V_Intel,
     SPV_V_NVIDIA, SPV_V_Qualcomm, SPV_V_SwiftShader,
     SPV_V_Unknown
   ]>;
 
 //===----------------------------------------------------------------------===//
 // SPIR-V extension definitions
 //===----------------------------------------------------------------------===//
 
 // Extensions known to the SPIR-V dialect.
 // https://github.com/KhronosGroup/SPIRV-Registry has the full list.
-def SPV_KHR_16bit_storage                    : StrEnumAttrCase<"SPV_KHR_16bit_storage">;
-def SPV_KHR_8bit_storage                     : StrEnumAttrCase<"SPV_KHR_8bit_storage">;
-def SPV_KHR_device_group                     : StrEnumAttrCase<"SPV_KHR_device_group">;
-def SPV_KHR_float_controls                   : StrEnumAttrCase<"SPV_KHR_float_controls">;
-def SPV_KHR_physical_storage_buffer          : StrEnumAttrCase<"SPV_KHR_physical_storage_buffer">;
-def SPV_KHR_multiview                        : StrEnumAttrCase<"SPV_KHR_multiview">;
-def SPV_KHR_no_integer_wrap_decoration       : StrEnumAttrCase<"SPV_KHR_no_integer_wrap_decoration">;
-def SPV_KHR_post_depth_coverage              : StrEnumAttrCase<"SPV_KHR_post_depth_coverage">;
-def SPV_KHR_shader_atomic_counter_ops        : StrEnumAttrCase<"SPV_KHR_shader_atomic_counter_ops">;
-def SPV_KHR_shader_ballot                    : StrEnumAttrCase<"SPV_KHR_shader_ballot">;
-def SPV_KHR_shader_clock                     : StrEnumAttrCase<"SPV_KHR_shader_clock">;
-def SPV_KHR_shader_draw_parameters           : StrEnumAttrCase<"SPV_KHR_shader_draw_parameters">;
-def SPV_KHR_storage_buffer_storage_class     : StrEnumAttrCase<"SPV_KHR_storage_buffer_storage_class">;
-def SPV_KHR_subgroup_vote                    : StrEnumAttrCase<"SPV_KHR_subgroup_vote">;
-def SPV_KHR_variable_pointers                : StrEnumAttrCase<"SPV_KHR_variable_pointers">;
-def SPV_KHR_vulkan_memory_model              : StrEnumAttrCase<"SPV_KHR_vulkan_memory_model">;
-def SPV_KHR_expect_assume                    : StrEnumAttrCase<"SPV_KHR_expect_assume">;
-def SPV_KHR_integer_dot_product              : StrEnumAttrCase<"SPV_KHR_integer_dot_product">;
-def SPV_KHR_bit_instructions                 : StrEnumAttrCase<"SPV_KHR_bit_instructions">;
-def SPV_KHR_fragment_shading_rate            : StrEnumAttrCase<"SPV_KHR_fragment_shading_rate">;
-def SPV_KHR_workgroup_memory_explicit_layout : StrEnumAttrCase<"SPV_KHR_workgroup_memory_explicit_layout">;
-def SPV_KHR_ray_query                        : StrEnumAttrCase<"SPV_KHR_ray_query">;
-def SPV_KHR_ray_tracing                      : StrEnumAttrCase<"SPV_KHR_ray_tracing">;
-def SPV_KHR_subgroup_uniform_control_flow    : StrEnumAttrCase<"SPV_KHR_subgroup_uniform_control_flow">;
-def SPV_KHR_linkonce_odr                     : StrEnumAttrCase<"SPV_KHR_linkonce_odr">;
-
-def SPV_EXT_demote_to_helper_invocation  : StrEnumAttrCase<"SPV_EXT_demote_to_helper_invocation">;
-def SPV_EXT_descriptor_indexing          : StrEnumAttrCase<"SPV_EXT_descriptor_indexing">;
-def SPV_EXT_fragment_fully_covered       : StrEnumAttrCase<"SPV_EXT_fragment_fully_covered">;
-def SPV_EXT_fragment_invocation_density  : StrEnumAttrCase<"SPV_EXT_fragment_invocation_density">;
-def SPV_EXT_fragment_shader_interlock    : StrEnumAttrCase<"SPV_EXT_fragment_shader_interlock">;
-def SPV_EXT_physical_storage_buffer      : StrEnumAttrCase<"SPV_EXT_physical_storage_buffer">;
-def SPV_EXT_shader_stencil_export        : StrEnumAttrCase<"SPV_EXT_shader_stencil_export">;
-def SPV_EXT_shader_viewport_index_layer  : StrEnumAttrCase<"SPV_EXT_shader_viewport_index_layer">;
-def SPV_EXT_shader_atomic_float_add      : StrEnumAttrCase<"SPV_EXT_shader_atomic_float_add">;
-def SPV_EXT_shader_atomic_float_min_max  : StrEnumAttrCase<"SPV_EXT_shader_atomic_float_min_max">;
-def SPV_EXT_shader_image_int64           : StrEnumAttrCase<"SPV_EXT_shader_image_int64">;
-def SPV_EXT_shader_atomic_float16_add    : StrEnumAttrCase<"SPV_EXT_shader_atomic_float16_add">;
-
-def SPV_AMD_gpu_shader_half_float_fetch      : StrEnumAttrCase<"SPV_AMD_gpu_shader_half_float_fetch">;
-def SPV_AMD_shader_ballot                    : StrEnumAttrCase<"SPV_AMD_shader_ballot">;
-def SPV_AMD_shader_explicit_vertex_parameter : StrEnumAttrCase<"SPV_AMD_shader_explicit_vertex_parameter">;
-def SPV_AMD_shader_fragment_mask             : StrEnumAttrCase<"SPV_AMD_shader_fragment_mask">;
-def SPV_AMD_shader_image_load_store_lod      : StrEnumAttrCase<"SPV_AMD_shader_image_load_store_lod">;
-def SPV_AMD_texture_gather_bias_lod          : StrEnumAttrCase<"SPV_AMD_texture_gather_bias_lod">;
-
-def SPV_GOOGLE_decorate_string           : StrEnumAttrCase<"SPV_GOOGLE_decorate_string">;
-def SPV_GOOGLE_hlsl_functionality1       : StrEnumAttrCase<"SPV_GOOGLE_hlsl_functionality1">;
-def SPV_GOOGLE_user_type                 : StrEnumAttrCase<"SPV_GOOGLE_user_type">;
-
-def SPV_INTEL_device_side_avc_motion_estimation  : StrEnumAttrCase<"SPV_INTEL_device_side_avc_motion_estimation">;
-def SPV_INTEL_media_block_io                     : StrEnumAttrCase<"SPV_INTEL_media_block_io">;
-def SPV_INTEL_shader_integer_functions2          : StrEnumAttrCase<"SPV_INTEL_shader_integer_functions2">;
-def SPV_INTEL_subgroups                          : StrEnumAttrCase<"SPV_INTEL_subgroups">;
-def SPV_INTEL_float_controls2                    : StrEnumAttrCase<"SPV_INTEL_float_controls2">;
-def SPV_INTEL_function_pointers                  : StrEnumAttrCase<"SPV_INTEL_function_pointers">;
-def SPV_INTEL_inline_assembly                    : StrEnumAttrCase<"SPV_INTEL_inline_assembly">;
-def SPV_INTEL_vector_compute                     : StrEnumAttrCase<"SPV_INTEL_vector_compute">;
-def SPV_INTEL_variable_length_array              : StrEnumAttrCase<"SPV_INTEL_variable_length_array">;
-def SPV_INTEL_fpga_memory_attributes             : StrEnumAttrCase<"SPV_INTEL_fpga_memory_attributes">;
-def SPV_INTEL_arbitrary_precision_integers       : StrEnumAttrCase<"SPV_INTEL_arbitrary_precision_integers">;
-def SPV_INTEL_arbitrary_precision_floating_point : StrEnumAttrCase<"SPV_INTEL_arbitrary_precision_floating_point">;
-def SPV_INTEL_unstructured_loop_controls         : StrEnumAttrCase<"SPV_INTEL_unstructured_loop_controls">;
-def SPV_INTEL_fpga_loop_controls                 : StrEnumAttrCase<"SPV_INTEL_fpga_loop_controls">;
-def SPV_INTEL_kernel_attributes                  : StrEnumAttrCase<"SPV_INTEL_kernel_attributes">;
-def SPV_INTEL_fpga_memory_accesses               : StrEnumAttrCase<"SPV_INTEL_fpga_memory_accesses">;
-def SPV_INTEL_fpga_cluster_attributes            : StrEnumAttrCase<"SPV_INTEL_fpga_cluster_attributes">;
-def SPV_INTEL_loop_fuse                          : StrEnumAttrCase<"SPV_INTEL_loop_fuse">;
-def SPV_INTEL_fpga_buffer_location               : StrEnumAttrCase<"SPV_INTEL_fpga_buffer_location">;
-def SPV_INTEL_arbitrary_precision_fixed_point    : StrEnumAttrCase<"SPV_INTEL_arbitrary_precision_fixed_point">;
-def SPV_INTEL_usm_storage_classes                : StrEnumAttrCase<"SPV_INTEL_usm_storage_classes">;
-def SPV_INTEL_io_pipes                           : StrEnumAttrCase<"SPV_INTEL_io_pipes">;
-def SPV_INTEL_blocking_pipes                     : StrEnumAttrCase<"SPV_INTEL_blocking_pipes">;
-def SPV_INTEL_fpga_reg                           : StrEnumAttrCase<"SPV_INTEL_fpga_reg">;
-def SPV_INTEL_long_constant_composite            : StrEnumAttrCase<"SPV_INTEL_long_constant_composite">;
-def SPV_INTEL_optnone                            : StrEnumAttrCase<"SPV_INTEL_optnone">;
-def SPV_INTEL_debug_module                       : StrEnumAttrCase<"SPV_INTEL_debug_module">;
-def SPV_INTEL_fp_fast_math_mode                  : StrEnumAttrCase<"SPV_INTEL_fp_fast_math_mode">;
-
-def SPV_NV_compute_shader_derivatives    : StrEnumAttrCase<"SPV_NV_compute_shader_derivatives">;
-def SPV_NV_cooperative_matrix            : StrEnumAttrCase<"SPV_NV_cooperative_matrix">;
-def SPV_NV_fragment_shader_barycentric   : StrEnumAttrCase<"SPV_NV_fragment_shader_barycentric">;
-def SPV_NV_geometry_shader_passthrough   : StrEnumAttrCase<"SPV_NV_geometry_shader_passthrough">;
-def SPV_NV_mesh_shader                   : StrEnumAttrCase<"SPV_NV_mesh_shader">;
-def SPV_NV_ray_tracing                   : StrEnumAttrCase<"SPV_NV_ray_tracing">;
-def SPV_NV_sample_mask_override_coverage : StrEnumAttrCase<"SPV_NV_sample_mask_override_coverage">;
-def SPV_NV_shader_image_footprint        : StrEnumAttrCase<"SPV_NV_shader_image_footprint">;
-def SPV_NV_shader_sm_builtins            : StrEnumAttrCase<"SPV_NV_shader_sm_builtins">;
-def SPV_NV_shader_subgroup_partitioned   : StrEnumAttrCase<"SPV_NV_shader_subgroup_partitioned">;
-def SPV_NV_shading_rate                  : StrEnumAttrCase<"SPV_NV_shading_rate">;
-def SPV_NV_stereo_view_rendering         : StrEnumAttrCase<"SPV_NV_stereo_view_rendering">;
-def SPV_NV_viewport_array2               : StrEnumAttrCase<"SPV_NV_viewport_array2">;
-def SPV_NV_bindless_texture              : StrEnumAttrCase<"SPV_NV_bindless_texture">;
-def SPV_NV_ray_tracing_motion_blur       : StrEnumAttrCase<"SPV_NV_ray_tracing_motion_blur">;
+def SPV_KHR_16bit_storage                    : I32EnumAttrCase<"SPV_KHR_16bit_storage", 0>;
+def SPV_KHR_8bit_storage                     : I32EnumAttrCase<"SPV_KHR_8bit_storage", 1>;
+def SPV_KHR_device_group                     : I32EnumAttrCase<"SPV_KHR_device_group", 2>;
+def SPV_KHR_float_controls                   : I32EnumAttrCase<"SPV_KHR_float_controls", 3>;
+def SPV_KHR_physical_storage_buffer          : I32EnumAttrCase<"SPV_KHR_physical_storage_buffer", 4>;
+def SPV_KHR_multiview                        : I32EnumAttrCase<"SPV_KHR_multiview", 5>;
+def SPV_KHR_no_integer_wrap_decoration       : I32EnumAttrCase<"SPV_KHR_no_integer_wrap_decoration", 6>;
+def SPV_KHR_post_depth_coverage              : I32EnumAttrCase<"SPV_KHR_post_depth_coverage", 7>;
+def SPV_KHR_shader_atomic_counter_ops        : I32EnumAttrCase<"SPV_KHR_shader_atomic_counter_ops", 8>;
+def SPV_KHR_shader_ballot                    : I32EnumAttrCase<"SPV_KHR_shader_ballot", 9>;
+def SPV_KHR_shader_clock                     : I32EnumAttrCase<"SPV_KHR_shader_clock", 10>;
+def SPV_KHR_shader_draw_parameters           : I32EnumAttrCase<"SPV_KHR_shader_draw_parameters", 11>;
+def SPV_KHR_storage_buffer_storage_class     : I32EnumAttrCase<"SPV_KHR_storage_buffer_storage_class", 12>;
+def SPV_KHR_subgroup_vote                    : I32EnumAttrCase<"SPV_KHR_subgroup_vote", 13>;
+def SPV_KHR_variable_pointers                : I32EnumAttrCase<"SPV_KHR_variable_pointers", 14>;
+def SPV_KHR_vulkan_memory_model              : I32EnumAttrCase<"SPV_KHR_vulkan_memory_model", 15>;
+def SPV_KHR_expect_assume                    : I32EnumAttrCase<"SPV_KHR_expect_assume", 16>;
+def SPV_KHR_integer_dot_product              : I32EnumAttrCase<"SPV_KHR_integer_dot_product", 17>;
+def SPV_KHR_bit_instructions                 : I32EnumAttrCase<"SPV_KHR_bit_instructions", 18>;
+def SPV_KHR_fragment_shading_rate            : I32EnumAttrCase<"SPV_KHR_fragment_shading_rate", 19>;
+def SPV_KHR_workgroup_memory_explicit_layout : I32EnumAttrCase<"SPV_KHR_workgroup_memory_explicit_layout", 20>;
+def SPV_KHR_ray_query                        : I32EnumAttrCase<"SPV_KHR_ray_query", 21>;
+def SPV_KHR_ray_tracing                      : I32EnumAttrCase<"SPV_KHR_ray_tracing", 22>;
+def SPV_KHR_subgroup_uniform_control_flow    : I32EnumAttrCase<"SPV_KHR_subgroup_uniform_control_flow", 23>;
+def SPV_KHR_linkonce_odr                     : I32EnumAttrCase<"SPV_KHR_linkonce_odr", 24>;
+
+def SPV_EXT_demote_to_helper_invocation  : I32EnumAttrCase<"SPV_EXT_demote_to_helper_invocation", 25>;
+def SPV_EXT_descriptor_indexing          : I32EnumAttrCase<"SPV_EXT_descriptor_indexing", 26>;
+def SPV_EXT_fragment_fully_covered       : I32EnumAttrCase<"SPV_EXT_fragment_fully_covered", 27>;
+def SPV_EXT_fragment_invocation_density  : I32EnumAttrCase<"SPV_EXT_fragment_invocation_density", 28>;
+def SPV_EXT_fragment_shader_interlock    : I32EnumAttrCase<"SPV_EXT_fragment_shader_interlock", 29>;
+def SPV_EXT_physical_storage_buffer      : I32EnumAttrCase<"SPV_EXT_physical_storage_buffer", 30>;
+def SPV_EXT_shader_stencil_export        : I32EnumAttrCase<"SPV_EXT_shader_stencil_export", 31>;
+def SPV_EXT_shader_viewport_index_layer  : I32EnumAttrCase<"SPV_EXT_shader_viewport_index_layer", 32>;
+def SPV_EXT_shader_atomic_float_add      : I32EnumAttrCase<"SPV_EXT_shader_atomic_float_add", 33>;
+def SPV_EXT_shader_atomic_float_min_max  : I32EnumAttrCase<"SPV_EXT_shader_atomic_float_min_max", 34>;
+def SPV_EXT_shader_image_int64           : I32EnumAttrCase<"SPV_EXT_shader_image_int64", 35>;
+def SPV_EXT_shader_atomic_float16_add    : I32EnumAttrCase<"SPV_EXT_shader_atomic_float16_add", 36>;
+
+def SPV_AMD_gpu_shader_half_float_fetch      : I32EnumAttrCase<"SPV_AMD_gpu_shader_half_float_fetch", 37>;
+def SPV_AMD_shader_ballot                    : I32EnumAttrCase<"SPV_AMD_shader_ballot", 38>;
+def SPV_AMD_shader_explicit_vertex_parameter : I32EnumAttrCase<"SPV_AMD_shader_explicit_vertex_parameter", 39>;
+def SPV_AMD_shader_fragment_mask             : I32EnumAttrCase<"SPV_AMD_shader_fragment_mask", 40>;
+def SPV_AMD_shader_image_load_store_lod      : I32EnumAttrCase<"SPV_AMD_shader_image_load_store_lod", 41>;
+def SPV_AMD_texture_gather_bias_lod          : I32EnumAttrCase<"SPV_AMD_texture_gather_bias_lod", 42>;
+
+def SPV_GOOGLE_decorate_string           : I32EnumAttrCase<"SPV_GOOGLE_decorate_string", 43>;
+def SPV_GOOGLE_hlsl_functionality1       : I32EnumAttrCase<"SPV_GOOGLE_hlsl_functionality1", 44>;
+def SPV_GOOGLE_user_type                 : I32EnumAttrCase<"SPV_GOOGLE_user_type", 45>;
+
+def SPV_INTEL_device_side_avc_motion_estimation  : I32EnumAttrCase<"SPV_INTEL_device_side_avc_motion_estimation", 46>;
+def SPV_INTEL_media_block_io                     : I32EnumAttrCase<"SPV_INTEL_media_block_io", 47>;
+def SPV_INTEL_shader_integer_functions2          : I32EnumAttrCase<"SPV_INTEL_shader_integer_functions2", 48>;
+def SPV_INTEL_subgroups                          : I32EnumAttrCase<"SPV_INTEL_subgroups", 49>;
+def SPV_INTEL_float_controls2                    : I32EnumAttrCase<"SPV_INTEL_float_controls2", 50>;
+def SPV_INTEL_function_pointers                  : I32EnumAttrCase<"SPV_INTEL_function_pointers", 51>;
+def SPV_INTEL_inline_assembly                    : I32EnumAttrCase<"SPV_INTEL_inline_assembly", 52>;
+def SPV_INTEL_vector_compute                     : I32EnumAttrCase<"SPV_INTEL_vector_compute", 53>;
+def SPV_INTEL_variable_length_array              : I32EnumAttrCase<"SPV_INTEL_variable_length_array", 54>;
+def SPV_INTEL_fpga_memory_attributes             : I32EnumAttrCase<"SPV_INTEL_fpga_memory_attributes", 55>;
+def SPV_INTEL_arbitrary_precision_integers       : I32EnumAttrCase<"SPV_INTEL_arbitrary_precision_integers", 56>;
+def SPV_INTEL_arbitrary_precision_floating_point : I32EnumAttrCase<"SPV_INTEL_arbitrary_precision_floating_point", 57>;
+def SPV_INTEL_unstructured_loop_controls         : I32EnumAttrCase<"SPV_INTEL_unstructured_loop_controls", 58>;
+def SPV_INTEL_fpga_loop_controls                 : I32EnumAttrCase<"SPV_INTEL_fpga_loop_controls", 59>;
+def SPV_INTEL_kernel_attributes                  : I32EnumAttrCase<"SPV_INTEL_kernel_attributes", 60>;
+def SPV_INTEL_fpga_memory_accesses               : I32EnumAttrCase<"SPV_INTEL_fpga_memory_accesses", 61>;
+def SPV_INTEL_fpga_cluster_attributes            : I32EnumAttrCase<"SPV_INTEL_fpga_cluster_attributes", 62>;
+def SPV_INTEL_loop_fuse                          : I32EnumAttrCase<"SPV_INTEL_loop_fuse", 63>;
+def SPV_INTEL_fpga_buffer_location               : I32EnumAttrCase<"SPV_INTEL_fpga_buffer_location", 64>;
+def SPV_INTEL_arbitrary_precision_fixed_point    : I32EnumAttrCase<"SPV_INTEL_arbitrary_precision_fixed_point", 65>;
+def SPV_INTEL_usm_storage_classes                : I32EnumAttrCase<"SPV_INTEL_usm_storage_classes", 66>;
+def SPV_INTEL_io_pipes                           : I32EnumAttrCase<"SPV_INTEL_io_pipes", 67>;
+def SPV_INTEL_blocking_pipes                     : I32EnumAttrCase<"SPV_INTEL_blocking_pipes", 68>;
+def SPV_INTEL_fpga_reg                           : I32EnumAttrCase<"SPV_INTEL_fpga_reg", 69>;
+def SPV_INTEL_long_constant_composite            : I32EnumAttrCase<"SPV_INTEL_long_constant_composite", 70>;
+def SPV_INTEL_optnone                            : I32EnumAttrCase<"SPV_INTEL_optnone", 71>;
+def SPV_INTEL_debug_module                       : I32EnumAttrCase<"SPV_INTEL_debug_module", 72>;
+def SPV_INTEL_fp_fast_math_mode                  : I32EnumAttrCase<"SPV_INTEL_fp_fast_math_mode", 73>;
+
+def SPV_NV_compute_shader_derivatives    : I32EnumAttrCase<"SPV_NV_compute_shader_derivatives", 74>;
+def SPV_NV_cooperative_matrix            : I32EnumAttrCase<"SPV_NV_cooperative_matrix", 75>;
+def SPV_NV_fragment_shader_barycentric   : I32EnumAttrCase<"SPV_NV_fragment_shader_barycentric", 76>;
+def SPV_NV_geometry_shader_passthrough   : I32EnumAttrCase<"SPV_NV_geometry_shader_passthrough", 77>;
+def SPV_NV_mesh_shader                   : I32EnumAttrCase<"SPV_NV_mesh_shader", 78>;
+def SPV_NV_ray_tracing                   : I32EnumAttrCase<"SPV_NV_ray_tracing", 79>;
+def SPV_NV_sample_mask_override_coverage : I32EnumAttrCase<"SPV_NV_sample_mask_override_coverage", 80>;
+def SPV_NV_shader_image_footprint        : I32EnumAttrCase<"SPV_NV_shader_image_footprint", 81>;
+def SPV_NV_shader_sm_builtins            : I32EnumAttrCase<"SPV_NV_shader_sm_builtins", 82>;
+def SPV_NV_shader_subgroup_partitioned   : I32EnumAttrCase<"SPV_NV_shader_subgroup_partitioned", 83>;
+def SPV_NV_shading_rate                  : I32EnumAttrCase<"SPV_NV_shading_rate", 84>;
+def SPV_NV_stereo_view_rendering         : I32EnumAttrCase<"SPV_NV_stereo_view_rendering", 85>;
+def SPV_NV_viewport_array2               : I32EnumAttrCase<"SPV_NV_viewport_array2", 86>;
+def SPV_NV_bindless_texture              : I32EnumAttrCase<"SPV_NV_bindless_texture", 87>;
+def SPV_NV_ray_tracing_motion_blur       : I32EnumAttrCase<"SPV_NV_ray_tracing_motion_blur", 88>;
 
-def SPV_NVX_multiview_per_view_attributes : StrEnumAttrCase<"SPV_NVX_multiview_per_view_attributes">;
+def SPV_NVX_multiview_per_view_attributes : I32EnumAttrCase<"SPV_NVX_multiview_per_view_attributes", 89>;
 
 def SPV_ExtensionAttr :
-    SPV_StrEnumAttr<"Extension", "supported SPIR-V extensions", [
+    SPV_EnumAttr<"Extension", "supported SPIR-V extensions", "ext", [
       SPV_KHR_16bit_storage, SPV_KHR_8bit_storage, SPV_KHR_device_group,
       SPV_KHR_float_controls, SPV_KHR_physical_storage_buffer, SPV_KHR_multiview,
       SPV_KHR_no_integer_wrap_decoration, SPV_KHR_post_depth_coverage,
       SPV_KHR_shader_atomic_counter_ops, SPV_KHR_shader_ballot,
       SPV_KHR_shader_clock, SPV_KHR_shader_draw_parameters,
       SPV_KHR_storage_buffer_storage_class, SPV_KHR_subgroup_vote,
       SPV_KHR_variable_pointers, SPV_KHR_vulkan_memory_model, SPV_KHR_expect_assume,
       SPV_KHR_integer_dot_product, SPV_KHR_bit_instructions, SPV_KHR_fragment_shading_rate,

mlir/include/mlir/Dialect/SPIRV/IR/SPIRVEnums.h

 #include "llvm/ADT/StringRef.h"
 
 // Forward declare enum classes related to op availability. Their definitions
 // are in the TableGen'erated SPIRVEnums.h.inc and can be referenced by other
 // declarations in SPIRVEnums.h.inc.
 namespace mlir {
 namespace spirv {
 enum class Version : uint32_t;
-enum class Extension;
+enum class Extension : uint32_t;
 enum class Capability : uint32_t;
 } // namespace spirv
 } // namespace mlir
 
 // Pull in all enum type definitions and utility function declarations
 #include "mlir/Dialect/SPIRV/IR/SPIRVEnums.h.inc"
 
 // Pull in all enum type availability query function declarations

mlir/lib/Conversion/GPUCommon/IndexIntrinsicsOpLowering.h

 // Rewriting that replaces Op with XOp, YOp, or ZOp depending on the dimension
 // that Op operates on.  Op is assumed to return an `std.index` value and
 // XOp, YOp and ZOp are assumed to return an `llvm.i32` value.  Depending on
 // `indexBitwidth`, sign-extend or truncate the resulting value to match the
 // bitwidth expected by the consumers of the value.
 template <typename Op, typename XOp, typename YOp, typename ZOp>
 struct GPUIndexIntrinsicOpLowering : public ConvertOpToLLVMPattern<Op> {
 private:
-  enum Dimension { X = 0, Y = 1, Z = 2, invalid };
   unsigned indexBitwidth;
 
-  static Dimension dimensionToIndex(Op op) {
-    return StringSwitch<Dimension>(op.dimension())
-        .Case("x", X)
-        .Case("y", Y)
-        .Case("z", Z)
-        .Default(invalid);
-  }
-
 public:
   explicit GPUIndexIntrinsicOpLowering(LLVMTypeConverter &typeConverter)
       : ConvertOpToLLVMPattern<Op>(typeConverter),
         indexBitwidth(typeConverter.getIndexTypeBitwidth()) {}
 
   // Convert the kernel arguments to an LLVM type, preserve the rest.
   LogicalResult
   matchAndRewrite(Op op, typename Op::Adaptor adaptor,
                   ConversionPatternRewriter &rewriter) const override {
     auto loc = op->getLoc();
     MLIRContext *context = rewriter.getContext();
     Value newOp;
-    switch (dimensionToIndex(op)) {
-    case X:
+    switch (op.dimension()) {
+    case gpu::Dimension::x:
       newOp = rewriter.create<XOp>(loc, IntegerType::get(context, 32));
       break;
-    case Y:
+    case gpu::Dimension::y:
       newOp = rewriter.create<YOp>(loc, IntegerType::get(context, 32));
       break;
-    case Z:
+    case gpu::Dimension::z:
       newOp = rewriter.create<ZOp>(loc, IntegerType::get(context, 32));
       break;
-    default:
-      return failure();
     }
 
     if (indexBitwidth > 32) {
       newOp = rewriter.create<LLVM::SExtOp>(
           loc, IntegerType::get(context, indexBitwidth), newOp);
     } else if (indexBitwidth < 32) {
       newOp = rewriter.create<LLVM::TruncOp>(
           loc, IntegerType::get(context, indexBitwidth), newOp);

mlir/lib/Conversion/GPUToNVVM/LowerGpuOpsToNVVMOps.cpp

 #include "../GPUCommon/OpToFuncCallLowering.h"
 #include "../PassDetail.h"
 
 using namespace mlir;
 
 namespace {
 
 /// Convert gpu dialect shfl mode enum to the equivalent nvvm one.
-static NVVM::ShflKind convertShflKind(gpu::ShuffleModeAttr mode) {
+static NVVM::ShflKind convertShflKind(gpu::ShuffleMode mode) {
   switch (mode) {
-  case gpu::ShuffleModeAttr::XOR:
+  case gpu::ShuffleMode::XOR:
     return NVVM::ShflKind::bfly;
-  case gpu::ShuffleModeAttr::UP:
+  case gpu::ShuffleMode::UP:
     return NVVM::ShflKind::up;
-  case gpu::ShuffleModeAttr::DOWN:
+  case gpu::ShuffleMode::DOWN:
     return NVVM::ShflKind::down;
-  case gpu::ShuffleModeAttr::IDX:
+  case gpu::ShuffleMode::IDX:
     return NVVM::ShflKind::idx;
   }
   llvm_unreachable("unknown shuffle mode");
 }
 
 struct GPUShuffleOpLowering : public ConvertOpToLLVMPattern<gpu::ShuffleOp> {
   using ConvertOpToLLVMPattern<gpu::ShuffleOp>::ConvertOpToLLVMPattern;
 

mlir/lib/Conversion/GPUToSPIRV/GPUToSPIRV.cpp

 };
 
 } // namespace
 
 //===----------------------------------------------------------------------===//
 // Builtins.
 //===----------------------------------------------------------------------===//
 
-static Optional<int32_t> getLaunchConfigIndex(Operation *op) {
-  auto dimAttr = op->getAttrOfType<StringAttr>("dimension");
-  if (!dimAttr)
-    return llvm::None;
-
-  return llvm::StringSwitch<Optional<int32_t>>(dimAttr.getValue())
-      .Case("x", 0)
-      .Case("y", 1)
-      .Case("z", 2)
-      .Default(llvm::None);
-}
-
 template <typename SourceOp, spirv::BuiltIn builtin>
 LogicalResult LaunchConfigConversion<SourceOp, builtin>::matchAndRewrite(
     SourceOp op, typename SourceOp::Adaptor adaptor,
     ConversionPatternRewriter &rewriter) const {
-  auto index = getLaunchConfigIndex(op);
-  if (!index)
-    return failure();
-
   auto *typeConverter = this->template getTypeConverter<SPIRVTypeConverter>();
   auto indexType = typeConverter->getIndexType();
 
   // SPIR-V invocation builtin variables are a vector of type <3xi32>
   auto spirvBuiltin =
       spirv::getBuiltinVariableValue(op, builtin, indexType, rewriter);
   rewriter.replaceOpWithNewOp<spirv::CompositeExtractOp>(
       op, indexType, spirvBuiltin,
-      rewriter.getI32ArrayAttr({index.getValue()}));
+      rewriter.getI32ArrayAttr({static_cast<int32_t>(op.dimension())}));
   return success();
 }
 
 template <typename SourceOp, spirv::BuiltIn builtin>
 LogicalResult
 SingleDimLaunchConfigConversion<SourceOp, builtin>::matchAndRewrite(
     SourceOp op, typename SourceOp::Adaptor adaptor,
     ConversionPatternRewriter &rewriter) const {
   auto *typeConverter = this->template getTypeConverter<SPIRVTypeConverter>();
   auto indexType = typeConverter->getIndexType();
 
   auto spirvBuiltin =
       spirv::getBuiltinVariableValue(op, builtin, indexType, rewriter);
   rewriter.replaceOp(op, spirvBuiltin);
   return success();
 }
 
 LogicalResult WorkGroupSizeConversion::matchAndRewrite(
     gpu::BlockDimOp op, OpAdaptor adaptor,
     ConversionPatternRewriter &rewriter) const {
-  auto index = getLaunchConfigIndex(op);
-  if (!index)
-    return failure();
-
   auto workGroupSizeAttr = spirv::lookupLocalWorkGroupSize(op);
-  auto val = workGroupSizeAttr.getValues<int32_t>()[index.getValue()];
+  auto val = workGroupSizeAttr
+                 .getValues<int32_t>()[static_cast<int32_t>(op.dimension())];
   auto convertedType =
       getTypeConverter()->convertType(op.getResult().getType());
   if (!convertedType)
     return failure();
   rewriter.replaceOpWithNewOp<spirv::ConstantOp>(
       op, convertedType, IntegerAttr::get(convertedType, val));
   return success();
 }

mlir/lib/Dialect/GPU/CMakeLists.txt

 add_mlir_dialect_library(MLIRGPUOps
   IR/GPUDialect.cpp
 
   ADDITIONAL_HEADER_DIRS
   ${MLIR_MAIN_INCLUDE_DIR}/mlir/Dialect/GPU
 
   DEPENDS
   MLIRGPUOpsIncGen
+  MLIRGPUOpsAttributesIncGen
   MLIRGPUOpsEnumsGen
   MLIRGPUOpInterfacesIncGen
 
   LINK_LIBS PUBLIC
   MLIRArithmetic
   MLIRDLTI
   MLIRIR
   MLIRMemRef

mlir/lib/Dialect/GPU/IR/GPUDialect.cpp

 
 void GPUDialect::initialize() {
   addTypes<AsyncTokenType>();
   addTypes<MMAMatrixType>();
   addOperations<
 #define GET_OP_LIST
 #include "mlir/Dialect/GPU/GPUOps.cpp.inc"
       >();
+  addAttributes<
+#define GET_ATTRDEF_LIST
+#include "mlir/Dialect/GPU/GPUOpsAttributes.cpp.inc"
+      >();
   addInterfaces<GPUInlinerInterface>();
 }
 
 Type GPUDialect::parseType(DialectAsmParser &parser) const {
   // Parse the main keyword for the type.
   StringRef keyword;
   if (parser.parseKeyword(&keyword))
     return Type();
for (Block &block : allReduce.body()) {
         if (yield.getOperand(0).getType() != allReduce.getType())
           return allReduce.emitError("incorrect gpu.yield type");
         ++yieldCount;
       }
     }
     if (yieldCount == 0)
       return allReduce.emitError("expected gpu.yield op in region");
   } else {
-    StringRef opName = *allReduce.op();
-    if ((opName == "and" || opName == "or" || opName == "xor") &&
+    gpu::AllReduceOperation opName = *allReduce.op();
+    if ((opName == gpu::AllReduceOperation::AND ||
+         opName == gpu::AllReduceOperation::OR ||
+         opName == gpu::AllReduceOperation::XOR) &&
         !allReduce.getType().isa<IntegerType>()) {
       return allReduce.emitError()
-             << '`' << opName << '`'
+             << '`' << gpu::stringifyAllReduceOperation(opName) << '`'
              << " accumulator is only compatible with Integer type";
     }
   }
   return success();
 }
 
-static LogicalResult verifyShuffleOp(gpu::ShuffleOp shuffleOp) {
-  auto type = shuffleOp.value().getType();
-  if (shuffleOp.result().getType() != type) {
-    return shuffleOp.emitOpError()
-           << "requires the same type for value operand and result";
-  }
-  if (!type.isSignlessIntOrFloat() || type.getIntOrFloatBitWidth() != 32) {
-    return shuffleOp.emitOpError()
-           << "requires value operand type to be f32 or i32";
+// TODO: Support optional custom attributes (without dialect prefix).
+static ParseResult parseAllReduceOperation(AsmParser &parser,
+                                           AllReduceOperationAttr &attr) {
+  StringRef enumStr;
+  if (!parser.parseOptionalKeyword(&enumStr)) {
+    Optional<AllReduceOperation> op = gpu::symbolizeAllReduceOperation(enumStr);
+    if (!op)
+      return parser.emitError(parser.getCurrentLocation(), "invalid op kind");
+    attr = AllReduceOperationAttr::get(parser.getContext(), *op);
   }
   return success();
 }
 
-static void printShuffleOp(OpAsmPrinter &p, ShuffleOp op) {
-  p << ' ' << op.getOperands() << ' ' << stringifyEnum(op.mode()) << " : "
-    << op.value().getType();
-}
-
-static ParseResult parseShuffleOp(OpAsmParser &parser, OperationState &state) {
-  SmallVector<OpAsmParser::OperandType, 3> operandInfo;
-  if (parser.parseOperandList(operandInfo, 3))
-    return failure();
-
-  StringRef mode;
-  if (parser.parseKeyword(&mode))
-    return failure();
-  state.addAttribute("mode", parser.getBuilder().getStringAttr(mode));
-
-  Type valueType;
-  Type int32Type = parser.getBuilder().getIntegerType(32);
-  Type int1Type = parser.getBuilder().getI1Type();
-  if (parser.parseColonType(valueType) ||
-      parser.resolveOperands(operandInfo, {valueType, int32Type, int32Type},
-                             parser.getCurrentLocation(), state.operands) ||
-      parser.addTypesToList({valueType, int1Type}, state.types))
-    return failure();
-  return success();
+static void printAllReduceOperation(AsmPrinter &printer, Operation *op,
+                                    AllReduceOperationAttr attr) {
+  if (attr)
+    attr.print(printer);
 }
 
 //===----------------------------------------------------------------------===//
 // AsyncOpInterface
 //===----------------------------------------------------------------------===//
 
 void gpu::addAsyncDependency(Operation *op, Value token) {
   op->insertOperands(0, {token});

 void AllocOp::getCanonicalizationPatterns(RewritePatternSet &results,
                                           MLIRContext *context) {
   results.add<SimplifyDimOfAllocOp>(context);
 }
 
 #include "mlir/Dialect/GPU/GPUOpInterfaces.cpp.inc"
 #include "mlir/Dialect/GPU/GPUOpsEnums.cpp.inc"
 
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/GPU/GPUOpsAttributes.cpp.inc"
+
 #define GET_OP_CLASSES
 #include "mlir/Dialect/GPU/GPUOps.cpp.inc"

mlir/lib/Dialect/GPU/Transforms/AllReduceLowering.cpp

struct GpuAllReduceRewriter {
   ///     gpu.barrier
   ///     %result = load %workgroup_buffer[%zero]
   ///     return %result
   ///
   void rewrite() {
     rewriter.setInsertionPoint(reduceOp);
 
     // Compute linear invocation index and workgroup size.
-    Value dimX = getDimOp<gpu::BlockDimOp>("x");
-    Value dimY = getDimOp<gpu::BlockDimOp>("y");
-    Value dimZ = getDimOp<gpu::BlockDimOp>("z");
-    Value tidX = getDimOp<gpu::ThreadIdOp>("x");
-    Value tidY = getDimOp<gpu::ThreadIdOp>("y");
-    Value tidZ = getDimOp<gpu::ThreadIdOp>("z");
+    Value dimX = getDimOp<gpu::BlockDimOp>(gpu::Dimension::x);
+    Value dimY = getDimOp<gpu::BlockDimOp>(gpu::Dimension::y);
+    Value dimZ = getDimOp<gpu::BlockDimOp>(gpu::Dimension::z);
+    Value tidX = getDimOp<gpu::ThreadIdOp>(gpu::Dimension::x);
+    Value tidY = getDimOp<gpu::ThreadIdOp>(gpu::Dimension::y);
+    Value tidZ = getDimOp<gpu::ThreadIdOp>(gpu::Dimension::z);
     Value tmp1 = create<arith::MulIOp>(int32Type, tidZ, dimY);
     Value tmp2 = create<arith::AddIOp>(int32Type, tmp1, tidY);
     Value tmp3 = create<arith::MulIOp>(int32Type, tmp2, dimX);
     Value tmp4 = create<arith::MulIOp>(int32Type, dimX, dimY);
     Value invocationIdx = create<arith::AddIOp>(int32Type, tmp3, tidX);
     Value workgroupSize = create<arith::MulIOp>(int32Type, tmp4, dimZ);
 
     // Compute lane id (invocation id withing the subgroup).
private:
   // Shortcut to create an op from rewriter using loc as the first argument.
   template <typename T, typename... Args>
   T create(Args... args) {
     return rewriter.create<T>(loc, std::forward<Args>(args)...);
   }
 
   // Creates dimension op of type T, with the result casted to int32.
   template <typename T>
-  Value getDimOp(StringRef dimension) {
-    Value dim = create<T>(indexType, rewriter.getStringAttr(dimension));
+  Value getDimOp(gpu::Dimension dimension) {
+    Value dim = create<T>(indexType, dimension);
     return create<arith::IndexCastOp>(int32Type, dim);
   }
 
   /// Adds type to funcOp's workgroup attributions.
   Value createWorkgroupBuffer() {
     int workgroupMemoryAddressSpace =
         gpu::GPUDialect::getWorkgroupAddressSpace();
     auto bufferType = MemRefType::get({kSubgroupSize}, valueType, AffineMap{},
return AccumulatorFactory([&](Value lhs, Value rhs) {
 
       // Return accumulator result.
       rewriter.setInsertionPointToStart(split);
       return split->addArgument(lhs.getType());
     });
   }
 
   /// Returns an accumulator factory that creates an op specified by opName.
-  AccumulatorFactory getFactory(StringRef opName) {
+  AccumulatorFactory getFactory(gpu::AllReduceOperation opName) {
     bool isFloatingPoint = valueType.isa<FloatType>();
-    if (opName == "add")
+    switch (opName) {
+    case gpu::AllReduceOperation::ADD:
       return isFloatingPoint ? getFactory<arith::AddFOp>()
                              : getFactory<arith::AddIOp>();
-    if (opName == "mul")
+    case gpu::AllReduceOperation::MUL:
       return isFloatingPoint ? getFactory<arith::MulFOp>()
                              : getFactory<arith::MulIOp>();
-    if (opName == "and") {
+    case gpu::AllReduceOperation::AND:
       return getFactory<arith::AndIOp>();
-    }
-    if (opName == "or") {
+    case gpu::AllReduceOperation::OR:
       return getFactory<arith::OrIOp>();
-    }
-    if (opName == "xor") {
+    case gpu::AllReduceOperation::XOR:
       return getFactory<arith::XOrIOp>();
-    }
-    if (opName == "max") {
+    case gpu::AllReduceOperation::MAX:
       return isFloatingPoint
                  ? getCmpFactory<arith::CmpFOp, arith::CmpFPredicate,
                                  arith::CmpFPredicate::UGT>()
                  : getCmpFactory<arith::CmpIOp, arith::CmpIPredicate,
                                  arith::CmpIPredicate::ugt>();
-    }
-    if (opName == "min") {
+    case gpu::AllReduceOperation::MIN:
       return isFloatingPoint
                  ? getCmpFactory<arith::CmpFOp, arith::CmpFPredicate,
                                  arith::CmpFPredicate::ULT>()
                  : getCmpFactory<arith::CmpIOp, arith::CmpIPredicate,
                                  arith::CmpIPredicate::ult>();
-    }
+    default:
       return AccumulatorFactory();
     }
+  }
 
   /// Returns an accumulator factory that creates an op of type T.
   template <typename T>
   AccumulatorFactory getFactory() {
     return [&](Value lhs, Value rhs) {
       return create<T>(lhs.getType(), lhs, rhs);
     };
   }
private:
   /// the entire subgroup if activeWidth is larger than the subgroup width.
   /// The first lane returns the result, all others return values are undefined.
   Value createSubgroupReduce(Value activeWidth, Value laneId, Value operand,
                              AccumulatorFactory &accumFactory) {
     Value subgroupSize = create<arith::ConstantIntOp>(kSubgroupSize, int32Type);
     Value isPartialSubgroup = create<arith::CmpIOp>(arith::CmpIPredicate::slt,
                                                     activeWidth, subgroupSize);
     std::array<Type, 2> shuffleType = {valueType, rewriter.getI1Type()};
-    auto xorAttr = rewriter.getStringAttr("xor");
 
     createIf(
         isPartialSubgroup,
         // Generate reduction over a (potentially) partial subgroup.
         [&] {
           Value value = operand;
           // Repeatedly shuffle value from 'laneId ^ i' and accumulate if source
           // lane is within the active range. The accumulated value is available
           // in the first lane.
           for (int i = 1; i < kSubgroupSize; i <<= 1) {
             Value offset = create<arith::ConstantIntOp>(i, int32Type);
-            auto shuffleOp = create<gpu::ShuffleOp>(shuffleType, value, offset,
-                                                    activeWidth, xorAttr);
+            auto shuffleOp = create<gpu::ShuffleOp>(
+                shuffleType, value, offset, activeWidth, gpu::ShuffleMode::XOR);
             // Skip the accumulation if the shuffle op read from a lane outside
             // of the active range.
             createIf(
                 shuffleOp.getResult(1),
                 [&] {
                   return SmallVector<Value, 1>{
                       accumFactory(value, shuffleOp.getResult(0))};
                 },
                 [&] { return llvm::makeArrayRef(value); });
             value = rewriter.getInsertionBlock()->getArgument(0);
           }
           return SmallVector<Value, 1>{value};
         },
         // Generate a reduction over the entire subgroup. This is a
         // specialization of the above reduction with unconditional
         // accumulation.
         [&] {
           Value value = operand;
           for (int i = 1; i < kSubgroupSize; i <<= 1) {
             Value offset = create<arith::ConstantIntOp>(i, int32Type);
-            auto shuffleOp = create<gpu::ShuffleOp>(shuffleType, value, offset,
-                                                    subgroupSize, xorAttr);
+            auto shuffleOp =
+                create<gpu::ShuffleOp>(shuffleType, value, offset, subgroupSize,
+                                       gpu::ShuffleMode::XOR);
             value = accumFactory(value, shuffleOp.getResult(0));
           }
           return SmallVector<Value, 1>{value};
         });
     return rewriter.getInsertionBlock()->getArgument(0);
   }
 
   /// Returns value divided by the subgroup size (i.e. 32).

mlir/lib/Dialect/GPU/Transforms/KernelOutlining.cpp

 #include "mlir/Support/LLVM.h"
 #include "mlir/Transforms/RegionUtils.h"
 
 using namespace mlir;
 
 template <typename OpTy>
 static void createForAllDimensions(OpBuilder &builder, Location loc,
                                    SmallVectorImpl<Value> &values) {
-  for (StringRef dim : {"x", "y", "z"}) {
-    Value v = builder.create<OpTy>(loc, builder.getIndexType(),
-                                   builder.getStringAttr(dim));
-    values.push_back(v);
-  }
+  for (auto dim : {gpu::Dimension::x, gpu::Dimension::y, gpu::Dimension::z})
+    values.push_back(builder.create<OpTy>(loc, builder.getIndexType(), dim));
 }
 
 /// Adds operations generating block/thread ids and grid/block dimensions at the
 /// beginning of the `launchFuncOpBody` region. Add mapping from argument in
 /// entry block of `launchOpBody`, to the corresponding result value of the
 /// added operations.
 static void injectGpuIndexOperations(Location loc, Region &launchFuncOpBody,
                                      Region &launchOpBody,

mlir/lib/Dialect/GPU/Transforms/MemoryPromotion.cpp

 #include "mlir/Dialect/StandardOps/IR/Ops.h"
 #include "mlir/IR/ImplicitLocOpBuilder.h"
 #include "mlir/Pass/Pass.h"
 #include "mlir/Transforms/LoopUtils.h"
 
 using namespace mlir;
 using namespace mlir::gpu;
 
-/// Returns the textual name of a GPU dimension.
-static StringRef getDimName(unsigned dim) {
-  if (dim == 0)
-    return "x";
-  if (dim == 1)
-    return "y";
-  if (dim == 2)
-    return "z";
-
-  llvm_unreachable("dimension ID overflow");
-}
-
 /// Emits the (imperfect) loop nest performing the copy between "from" and "to"
 /// values using the bounds derived from the "from" value. Emits at least
 /// GPUDialect::getNumWorkgroupDimensions() loops, completing the nest with
 /// single-iteration loops. Maps the innermost loops to thread dimensions, in
 /// reverse order to enable access coalescing in the innermost loop.
 static void insertCopyLoops(ImplicitLocOpBuilder &b, Value from, Value to) {
   auto memRefType = from.getType().cast<MemRefType>();
   auto rank = memRefType.getRank();
for (auto idx = 0; idx < rank; ++idx) {
     ubs.push_back(b.createOrFold<memref::DimOp>(
         from, b.create<arith::ConstantIndexOp>(idx)));
     steps.push_back(one);
   }
 
   // Obtain thread identifiers and block sizes, necessary to map to them.
   auto indexType = b.getIndexType();
   SmallVector<Value, 3> threadIds, blockDims;
-  for (unsigned i = 0; i < 3; ++i) {
-    auto dimName = b.getStringAttr(getDimName(i));
-    threadIds.push_back(b.create<gpu::ThreadIdOp>(indexType, dimName));
-    blockDims.push_back(b.create<gpu::BlockDimOp>(indexType, dimName));
+  for (auto dim : {gpu::Dimension::x, gpu::Dimension::y, gpu::Dimension::z}) {
+    threadIds.push_back(b.create<gpu::ThreadIdOp>(indexType, dim));
+    blockDims.push_back(b.create<gpu::BlockDimOp>(indexType, dim));
   }
 
   // Produce the loop nest with copies.
   SmallVector<Value, 8> ivs(lbs.size());
   mlir::scf::buildLoopNest(
       b, b.getLoc(), lbs, ubs, steps,
       [&](OpBuilder &b, Location loc, ValueRange loopIvs) {
         ivs.assign(loopIvs.begin(), loopIvs.end());

mlir/lib/Dialect/LLVMIR/IR/NVVMDialect.cpp

 // LLVM dialect.
 //
 //===----------------------------------------------------------------------===//
 
 #include "mlir/Dialect/LLVMIR/NVVMDialect.h"
 
 #include "mlir/IR/Builders.h"
 #include "mlir/IR/BuiltinTypes.h"
+#include "mlir/IR/DialectImplementation.h"
 #include "mlir/IR/MLIRContext.h"
 #include "mlir/IR/Operation.h"
 #include "mlir/IR/OperationSupport.h"
+#include "llvm/ADT/TypeSwitch.h"
 #include "llvm/AsmParser/Parser.h"
 #include "llvm/IR/Attributes.h"
 #include "llvm/IR/Function.h"
 #include "llvm/IR/Type.h"
 #include "llvm/Support/SourceMgr.h"
 
 using namespace mlir;
 using namespace NVVM;

 //===----------------------------------------------------------------------===//
 
 // TODO: This should be the llvm.nvvm dialect once this is supported.
 void NVVMDialect::initialize() {
   addOperations<
 #define GET_OP_LIST
 #include "mlir/Dialect/LLVMIR/NVVMOps.cpp.inc"
       >();
+  addAttributes<
+#define GET_ATTRDEF_LIST
+#include "mlir/Dialect/LLVMIR/NVVMOpsAttributes.cpp.inc"
+      >();
 
   // Support unknown operations because not all NVVM operations are
   // registered.
   allowUnknownOperations();
 }
 
 LogicalResult NVVMDialect::verifyOperationAttribute(Operation *op,
                                                     NamedAttribute attr) {
   // Kernel function attribute should be attached to functions.
   if (attr.getName() == NVVMDialect::getKernelFuncAttrName()) {
     if (!isa<LLVM::LLVMFuncOp>(op)) {
       return op->emitError() << "'" << NVVMDialect::getKernelFuncAttrName()
                              << "' attribute attached to unexpected op";
     }
   }
   return success();
 }
 
 #define GET_OP_CLASSES
 #include "mlir/Dialect/LLVMIR/NVVMOps.cpp.inc"
+
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/LLVMIR/NVVMOpsAttributes.cpp.inc"

mlir/lib/Dialect/OpenACC/IR/OpenACC.cpp

 //===- OpenACC.cpp - OpenACC MLIR Operations ------------------------------===//
 //
 // Part of the MLIR Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 // =============================================================================
 
 #include "mlir/Dialect/OpenACC/OpenACC.h"
 #include "mlir/Dialect/Arithmetic/IR/Arithmetic.h"
 #include "mlir/Dialect/OpenACC/OpenACCOpsEnums.cpp.inc"
 #include "mlir/Dialect/StandardOps/IR/Ops.h"
 #include "mlir/IR/Builders.h"
 #include "mlir/IR/BuiltinTypes.h"
+#include "mlir/IR/DialectImplementation.h"
 #include "mlir/IR/OpImplementation.h"
 #include "mlir/Transforms/DialectConversion.h"
+#include "llvm/ADT/TypeSwitch.h"
 
 using namespace mlir;
 using namespace acc;
 
 #include "mlir/Dialect/OpenACC/OpenACCOpsDialect.cpp.inc"
 
 //===----------------------------------------------------------------------===//
 // OpenACC operations
 //===----------------------------------------------------------------------===//
 
 void OpenACCDialect::initialize() {
   addOperations<
 #define GET_OP_LIST
 #include "mlir/Dialect/OpenACC/OpenACCOps.cpp.inc"
       >();
+  addAttributes<
+#define GET_ATTRDEF_LIST
+#include "mlir/Dialect/OpenACC/OpenACCOpsAttributes.cpp.inc"
+      >();
 }
 
 template <typename StructureOp>
 static ParseResult parseRegions(OpAsmParser &parser, OperationState &state,
                                 unsigned nRegions = 1) {
 
   SmallVector<Region *, 2> regions;
   for (unsigned i = 0; i < nRegions; ++i)
static LogicalResult verify(acc::WaitOp waitOp) {
   if (waitOp.waitDevnum() && waitOp.waitOperands().empty())
     return waitOp.emitError("wait_devnum cannot appear without waitOperands");
 
   return success();
 }
 
 #define GET_OP_CLASSES
 #include "mlir/Dialect/OpenACC/OpenACCOps.cpp.inc"
+
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/OpenACC/OpenACCOpsAttributes.cpp.inc"

mlir/lib/Dialect/OpenMP/IR/OpenMPDialect.cpp

 //===- OpenMPDialect.cpp - MLIR Dialect for OpenMP implementation ---------===//
 //
 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
 // See https://llvm.org/LICENSE.txt for license information.
 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
 //
 //===----------------------------------------------------------------------===//
 //
 // This file implements the OpenMP dialect and its operations.
 //
 //===----------------------------------------------------------------------===//
 
 #include "mlir/Dialect/OpenMP/OpenMPDialect.h"
 #include "mlir/Dialect/LLVMIR/LLVMTypes.h"
 #include "mlir/Dialect/StandardOps/IR/Ops.h"
 #include "mlir/IR/Attributes.h"
+#include "mlir/IR/DialectImplementation.h"
 #include "mlir/IR/OpImplementation.h"
 #include "mlir/IR/OperationSupport.h"
 
 #include "llvm/ADT/BitVector.h"
 #include "llvm/ADT/SmallString.h"
 #include "llvm/ADT/StringExtras.h"
 #include "llvm/ADT/StringRef.h"
 #include "llvm/ADT/StringSwitch.h"
+#include "llvm/ADT/TypeSwitch.h"
 #include <cstddef>
 
 #include "mlir/Dialect/OpenMP/OpenMPOpsDialect.cpp.inc"
 #include "mlir/Dialect/OpenMP/OpenMPOpsEnums.cpp.inc"
 #include "mlir/Dialect/OpenMP/OpenMPTypeInterfaces.cpp.inc"
 
 using namespace mlir;
 using namespace mlir::omp;

 };
 } // namespace
 
 void OpenMPDialect::initialize() {
   addOperations<
 #define GET_OP_LIST
 #include "mlir/Dialect/OpenMP/OpenMPOps.cpp.inc"
       >();
+  addAttributes<
+#define GET_ATTRDEF_LIST
+#include "mlir/Dialect/OpenMP/OpenMPOpsAttributes.cpp.inc"
+      >();
 
   LLVM::LLVMPointerType::attachInterface<
       PointerLikeModel<LLVM::LLVMPointerType>>(*getContext());
   MemRefType::attachInterface<PointerLikeModel<MemRefType>>(*getContext());
 }
 
 //===----------------------------------------------------------------------===//
 // ParallelOp
static void printParallelOp(OpAsmPrinter &p, ParallelOp op) {
   printDataVars(p, op.firstprivate_vars(), "firstprivate");
   printDataVars(p, op.shared_vars(), "shared");
   printDataVars(p, op.copyin_vars(), "copyin");
 
   if (!op.allocate_vars().empty())
     printAllocateAndAllocator(p, op.allocate_vars(), op.allocators_vars());
 
   if (auto def = op.default_val())
-    p << "default(" << def->drop_front(3) << ") ";
+    p << "default(" << stringifyClauseDefault(*def).drop_front(3) << ") ";
 
   if (auto bind = op.proc_bind_val())
     p << "proc_bind(" << bind << ") ";
 
   p.printRegion(op.getRegion());
 }
 
 //===----------------------------------------------------------------------===//
parseScheduleClause(OpAsmParser &parser, SmallString<8> &schedule,
   if (verifyScheduleModifiers(parser, modifiers))
     return failure();
 
   return success();
 }
 
 /// Print schedule clause
 static void printScheduleClause(OpAsmPrinter &p, StringRef &sched,
-                                llvm::Optional<StringRef> modifier, bool simd,
+                                Optional<ScheduleModifier> modifier, bool simd,
                                 Value scheduleChunkVar) {
   std::string schedLower = sched.lower();
   p << "schedule(" << schedLower;
   if (scheduleChunkVar)
     p << " = " << scheduleChunkVar;
-  if (modifier && modifier.hasValue())
-    p << ", " << modifier;
+  if (modifier)
+    p << ", " << stringifyScheduleModifier(*modifier);
   if (simd)
     p << ", simd";
   p << ") ";
 }
 
 //===----------------------------------------------------------------------===//
 // Parser, printer and verifier for ReductionVarList
 //===----------------------------------------------------------------------===//

 };
 
 //===----------------------------------------------------------------------===//
 // Parser for Clause List
 //===----------------------------------------------------------------------===//
 
 /// Parse a list of clauses. The clauses can appear in any order, but their
 /// operand segment indices are in the same order that they are passed in the
 /// `clauses` list. The operand segments are added over the prevSegments
 
 /// clause-list ::= clause clause-list | empty
 /// clause ::= if | num-threads | private | firstprivate | lastprivate |
 ///            shared | copyin | allocate | default | proc-bind | reduction |
 ///            nowait | linear | schedule | collapse | order | ordered |
 ///            inclusive
 /// if ::= `if` `(` ssa-id-and-type `)`
 /// num-threads ::= `num_threads` `(` ssa-id-and-type `)`
if (clauseKeyword == "if") {
       if (checkAllowed(allocateClause) ||
           parseAllocateAndAllocator(parser, allocates, allocateTypes,
                                     allocators, allocatorTypes))
         return failure();
       clauseSegments[pos[allocateClause]] = allocates.size();
       clauseSegments[pos[allocateClause] + 1] = allocators.size();
     } else if (clauseKeyword == "default") {
       StringRef defval;
+      llvm::SMLoc loc = parser.getCurrentLocation();
       if (checkAllowed(defaultClause) || parser.parseLParen() ||
           parser.parseKeyword(&defval) || parser.parseRParen())
         return failure();
       // The def prefix is required for the attribute as "private" is a keyword
       // in C++.
-      auto attr = parser.getBuilder().getStringAttr("def" + defval);
-      result.addAttribute("default_val", attr);
+      if (Optional<ClauseDefault> def =
+              symbolizeClauseDefault(("def" + defval).str())) {
+        result.addAttribute("default_val",
+                            ClauseDefaultAttr::get(parser.getContext(), *def));
+      } else {
+        return parser.emitError(loc, "invalid default clause");
+      }
     } else if (clauseKeyword == "proc_bind") {
       StringRef bind;
       if (checkAllowed(procBindClause) || parser.parseLParen() ||
           parser.parseKeyword(&bind) || parser.parseRParen())
         return failure();
       auto attr = parser.getBuilder().getStringAttr(bind);
       result.addAttribute("proc_bind_val", attr);
     } else if (clauseKeyword == "reduction") {
static ParseResult parseClauses(OpAsmParser &parser, OperationState &result,
   }
 
   // Add schedule parameters
   if (done[scheduleClause] && !schedule.empty()) {
     schedule[0] = llvm::toUpper(schedule[0]);
     auto attr = parser.getBuilder().getStringAttr(schedule);
     result.addAttribute("schedule_val", attr);
     if (!modifiers.empty()) {
-      auto mod = parser.getBuilder().getStringAttr(modifiers[0]);
-      result.addAttribute("schedule_modifier", mod);
+      llvm::SMLoc loc = parser.getCurrentLocation();
+      if (Optional<ScheduleModifier> mod =
+              symbolizeScheduleModifier(modifiers[0])) {
+        result.addAttribute(
+            "schedule_modifier",
+            ScheduleModifierAttr::get(parser.getContext(), *mod));
+      } else {
+        return parser.emitError(loc, "invalid schedule modifier");
+      }
       // Only SIMD attribute is allowed here!
       if (modifiers.size() > 1) {
         assert(symbolizeScheduleModifier(modifiers[1]) ==
                mlir::omp::ScheduleModifier::simd);
         auto attr = UnitAttr::get(parser.getBuilder().getContext());
         result.addAttribute("simd_modifier", attr);
       }
     }
if (op.memory_order()) {
     StringRef memoryOrder = op.memory_order().getValue();
     if (memoryOrder.equals("acq_rel") || memoryOrder.equals("acquire"))
       return op.emitError(
           "memory-order must not be acq_rel or acquire for atomic updates");
   }
   return success();
 }
 
+#define GET_ATTRDEF_CLASSES
+#include "mlir/Dialect/OpenMP/OpenMPOpsAttributes.cpp.inc"
+
 #define GET_OP_CLASSES
 #include "mlir/Dialect/OpenMP/OpenMPOps.cpp.inc"

mlir/lib/Target/LLVMIR/Dialect/OpenMP/OpenMPToLLVMIRTranslation.cpp

static void convertOmpOpRegions(
 
   // Remove the blocks and values defined in this region from the mapping since
   // they are not visible outside of this region. This allows the same region to
   // be converted several times, that is cloned, without clashes, and slightly
   // speeds up the lookups.
   moduleTranslation.forgetMapping(region);
 }
 
 /// Converts the OpenMP parallel operation to LLVM IR.
 static LogicalResult
 convertOmpParallel(Operation &opInst, llvm::IRBuilderBase &builder,
                    LLVM::ModuleTranslation &moduleTranslation) {
   using InsertPointTy = llvm::OpenMPIRBuilder::InsertPointTy;
   // TODO: support error propagation in OpenMPIRBuilder and use it instead of
   // relying on captured variables.
   LogicalResult bodyGenStatus = success();
 

 }
 
 /// Converts an OpenMP 'ordered' operation into LLVM IR using OpenMPIRBuilder.
 static LogicalResult
 convertOmpOrdered(Operation &opInst, llvm::IRBuilderBase &builder,
                   LLVM::ModuleTranslation &moduleTranslation) {
   auto orderedOp = cast<omp::OrderedOp>(opInst);
 
-  omp::ClauseDepend dependType =
-      *omp::symbolizeClauseDepend(orderedOp.depend_type_valAttr().getValue());
+  omp::ClauseDepend dependType = *orderedOp.depend_type_val();
   bool isDependSource = dependType == omp::ClauseDepend::dependsource;
   unsigned numLoops = orderedOp.num_loops_val().getValue();
   SmallVector<llvm::Value *> vecValues =
       moduleTranslation.lookupValues(orderedOp.depend_vec_vars());
 
   llvm::OpenMPIRBuilder::LocationDescription ompLoc(
       builder.saveIP(), builder.getCurrentDebugLocation());
   size_t indexVecValues = 0;
case omp::ClauseScheduleKind::Runtime:
       else
         schedType = llvm::omp::OMPScheduleType::Runtime;
       break;
     default:
       llvm_unreachable("Unknown schedule value");
       break;
     }
 
-    if (loop.schedule_modifier().hasValue()) {
-      omp::ScheduleModifier modifier =
-          *omp::symbolizeScheduleModifier(loop.schedule_modifier().getValue());
-      switch (modifier) {
+    if (Optional<omp::ScheduleModifier> modifier = loop.schedule_modifier()) {
+      switch (*modifier) {
       case omp::ScheduleModifier::monotonic:
         schedType |= llvm::omp::OMPScheduleType::ModifierMonotonic;
         break;
       case omp::ScheduleModifier::nonmonotonic:
         schedType |= llvm::omp::OMPScheduleType::ModifierNonmonotonic;
         break;
       default:
         // Nothing to do here.

mlir/test/Conversion/GPUToNVVM/gpu-to-nvvm.mlir

 // RUN: mlir-opt %s -convert-gpu-to-nvvm -split-input-file | FileCheck %s
 // RUN: mlir-opt %s -convert-gpu-to-nvvm='index-bitwidth=32' -split-input-file | FileCheck --check-prefix=CHECK32 %s
 
 gpu.module @test_module {
   // CHECK-LABEL: func @gpu_index_ops()
   // CHECK32-LABEL: func @gpu_index_ops()
   builtin.func @gpu_index_ops()
       -> (index, index, index, index, index, index,
           index, index, index, index, index, index) {
     // CHECK32-NOT: = llvm.sext %{{.*}} : i32 to i64
 
     // CHECK: = nvvm.read.ptx.sreg.tid.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %tIdX = "gpu.thread_id"() {dimension = "x"} : () -> (index)
+    %tIdX = gpu.thread_id x
     // CHECK: = nvvm.read.ptx.sreg.tid.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %tIdY = "gpu.thread_id"() {dimension = "y"} : () -> (index)
+    %tIdY = gpu.thread_id y
     // CHECK: = nvvm.read.ptx.sreg.tid.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %tIdZ = "gpu.thread_id"() {dimension = "z"} : () -> (index)
+    %tIdZ = gpu.thread_id z
 
     // CHECK: = nvvm.read.ptx.sreg.ntid.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bDimX = "gpu.block_dim"() {dimension = "x"} : () -> (index)
+    %bDimX = gpu.block_dim x
     // CHECK: = nvvm.read.ptx.sreg.ntid.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bDimY = "gpu.block_dim"() {dimension = "y"} : () -> (index)
+    %bDimY = gpu.block_dim y
     // CHECK: = nvvm.read.ptx.sreg.ntid.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bDimZ = "gpu.block_dim"() {dimension = "z"} : () -> (index)
+    %bDimZ = gpu.block_dim z
 
     // CHECK: = nvvm.read.ptx.sreg.ctaid.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bIdX = "gpu.block_id"() {dimension = "x"} : () -> (index)
+    %bIdX = gpu.block_id x
     // CHECK: = nvvm.read.ptx.sreg.ctaid.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bIdY = "gpu.block_id"() {dimension = "y"} : () -> (index)
+    %bIdY = gpu.block_id y
     // CHECK: = nvvm.read.ptx.sreg.ctaid.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bIdZ = "gpu.block_id"() {dimension = "z"} : () -> (index)
+    %bIdZ = gpu.block_id z
 
     // CHECK: = nvvm.read.ptx.sreg.nctaid.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %gDimX = "gpu.grid_dim"() {dimension = "x"} : () -> (index)
+    %gDimX = gpu.grid_dim x
     // CHECK: = nvvm.read.ptx.sreg.nctaid.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %gDimY = "gpu.grid_dim"() {dimension = "y"} : () -> (index)
+    %gDimY = gpu.grid_dim y
     // CHECK: = nvvm.read.ptx.sreg.nctaid.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %gDimZ = "gpu.grid_dim"() {dimension = "z"} : () -> (index)
+    %gDimZ = gpu.grid_dim z
 
     std.return %tIdX, %tIdY, %tIdZ, %bDimX, %bDimY, %bDimZ,
                %bIdX, %bIdY, %bIdZ, %gDimX, %gDimY, %gDimZ
         : index, index, index, index, index, index,
           index, index, index, index, index, index
   }
 }
 

 
 // -----
 
 gpu.module @test_module {
   // CHECK-LABEL: func @gpu_all_reduce_op()
   gpu.func @gpu_all_reduce_op() {
     %arg0 = arith.constant 1.0 : f32
     // TODO: Check full IR expansion once lowering has settled.
-    // CHECK: nvvm.shfl.sync "bfly" {{.*}}
+    // CHECK: nvvm.shfl.sync bfly {{.*}}
     // CHECK: nvvm.barrier0
     // CHECK: llvm.fadd
-    %result = "gpu.all_reduce"(%arg0) ({}) {op = "add"} : (f32) -> (f32)
+    %result = gpu.all_reduce add %arg0 {} : (f32) -> (f32)
 
     gpu.return
   }
 }
 
 // -----
 
 gpu.module @test_module {
   // CHECK-LABEL: func @gpu_all_reduce_region()
   gpu.func @gpu_all_reduce_region() {
     %arg0 = arith.constant 1 : i32
     // TODO: Check full IR expansion once lowering has settled.
-    // CHECK: nvvm.shfl.sync "bfly" {{.*}}
+    // CHECK: nvvm.shfl.sync bfly {{.*}}
     // CHECK: nvvm.barrier0
-    %result = "gpu.all_reduce"(%arg0) ({
+    %result = gpu.all_reduce %arg0 {
     ^bb(%lhs : i32, %rhs : i32):
       %xor = arith.xori %lhs, %rhs : i32
       "gpu.yield"(%xor) : (i32) -> ()
-    }) : (i32) -> (i32)
+    } : (i32) -> (i32)
     gpu.return
   }
 }
 
 // -----
 
 gpu.module @test_module {
   // CHECK-LABEL: func @gpu_shuffle()
   builtin.func @gpu_shuffle() -> (f32, f32, f32, f32) {
     // CHECK: %[[#VALUE:]] = llvm.mlir.constant(1.000000e+00 : f32) : f32
     %arg0 = arith.constant 1.0 : f32
     // CHECK: %[[#OFFSET:]] = llvm.mlir.constant(4 : i32) : i32
     %arg1 = arith.constant 4 : i32
     // CHECK: %[[#WIDTH:]] = llvm.mlir.constant(23 : i32) : i32
     %arg2 = arith.constant 23 : i32
     // CHECK: %[[#ONE:]] = llvm.mlir.constant(1 : i32) : i32
     // CHECK: %[[#SHL:]] = llvm.shl %[[#ONE]], %[[#WIDTH]] : i32
     // CHECK: %[[#MASK:]] = llvm.sub %[[#SHL]], %[[#ONE]] : i32
     // CHECK: %[[#CLAMP:]] = llvm.sub %[[#WIDTH]], %[[#ONE]] : i32
-    // CHECK: %[[#SHFL:]] = nvvm.shfl.sync "bfly" %[[#MASK]], %[[#VALUE]], %[[#OFFSET]], %[[#CLAMP]] {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+    // CHECK: %[[#SHFL:]] = nvvm.shfl.sync bfly %[[#MASK]], %[[#VALUE]], %[[#OFFSET]], %[[#CLAMP]] {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
     // CHECK: llvm.extractvalue %[[#SHFL]][0 : index] : !llvm.struct<(f32, i1)>
     // CHECK: llvm.extractvalue %[[#SHFL]][1 : index] : !llvm.struct<(f32, i1)>
-    %shfl, %pred = "gpu.shuffle"(%arg0, %arg1, %arg2) { mode = "xor" } : (f32, i32, i32) -> (f32, i1)
-    // CHECK: nvvm.shfl.sync "up" {{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
-    %shflu, %predu = "gpu.shuffle"(%arg0, %arg1, %arg2) { mode = "up" } : (f32, i32, i32) -> (f32, i1)
-    // CHECK: nvvm.shfl.sync "down" {{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
-    %shfld, %predd = "gpu.shuffle"(%arg0, %arg1, %arg2) { mode = "down" } : (f32, i32, i32) -> (f32, i1)
-    // CHECK: nvvm.shfl.sync "idx" {{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
-    %shfli, %predi = "gpu.shuffle"(%arg0, %arg1, %arg2) { mode = "idx" } : (f32, i32, i32) -> (f32, i1)
+    %shfl, %pred = gpu.shuffle xor %arg0, %arg1, %arg2 : f32
+    // CHECK: nvvm.shfl.sync up {{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+    %shflu, %predu = gpu.shuffle up %arg0, %arg1, %arg2 : f32
+    // CHECK: nvvm.shfl.sync down {{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+    %shfld, %predd = gpu.shuffle down %arg0, %arg1, %arg2 : f32
+    // CHECK: nvvm.shfl.sync idx {{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+    %shfli, %predi = gpu.shuffle idx %arg0, %arg1, %arg2 : f32
 
     std.return %shfl, %shflu, %shfld, %shfli : f32, f32,f32, f32
   }
 }
 
 // -----
 
 gpu.module @test_module {

mlir/test/Conversion/GPUToNVVM/wmma-ops-to-nvvm.mlir

builtin.func @gpu_wmma_load_op() -> (!gpu.mma_matrix<16x16xf16, "AOp">) {
     // CHECK: %{{.*}} = llvm.insertvalue %{{.*}}, %{{.*}}[{{.*}}, {{.*}}]
     // CHECK:  %[[BASE:.*]] = llvm.extractvalue %{{.*}}[1] : !llvm.struct<(ptr<f16, 3>, ptr<f16, 3>, i64, array<2 x i64>, array<2 x i64>)>
     // CHECK:  %[[LDM:.*]] = llvm.mlir.constant(32 : index) : i64
     // CHECK:  %[[LI:.*]] = llvm.mul %[[INX]], %[[LDM]]  : i64
     // CHECK:  %[[LIJ:.*]] = llvm.add %[[LI]], %[[INX]]  : i64
     // CHECK:  %[[ADDRESS:.*]] = llvm.getelementptr %[[BASE]][%[[LIJ]]] : (!llvm.ptr<f16, 3>, i64) -> !llvm.ptr<f16, 3>
     // CHECK:  %[[LDM32:.*]] = llvm.mlir.constant(32 : index) : i32
     // CHECK:  %[[FRAG:.*]] = nvvm.wmma.load %[[ADDRESS]], %[[LDM32]]
-    // CHECK-SAME: {eltype = "f16", frag = "a", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}  : (!llvm.ptr<f16, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
+    // CHECK-SAME: {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<a>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}  : (!llvm.ptr<f16, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  llvm.return %[[FRAG]] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 
     // CHECK32:  %[[INX:.*]] = llvm.mlir.constant(16 : index) : i32
     // CHECK32: %{{.*}} = llvm.insertvalue %{{.*}}, %{{.*}}[{{.*}}, {{.*}}]
     // CHECK32:  %[[BASE:.*]] = llvm.extractvalue %{{.*}}[1] : !llvm.struct<(ptr<f16, 3>, ptr<f16, 3>, i32, array<2 x i32>, array<2 x i32>)>
     // CHECK32:  %[[LDM:.*]] = llvm.mlir.constant(32 : index) : i32
     // CHECK32:  %[[LI:.*]] = llvm.mul %[[INX]], %[[LDM]]  : i32
     // CHECK32:  %[[LIJ:.*]] = llvm.add %[[LI]], %[[INX]]  : i32
     // CHECK32:  %[[ADDRESS:.*]] = llvm.getelementptr %[[BASE]][%[[LIJ]]] : (!llvm.ptr<f16, 3>, i32) -> !llvm.ptr<f16, 3>
     // CHECK32:  %[[LDM32:.*]] = llvm.mlir.constant(32 : index) : i32
     // CHECK32:  %[[FRAG:.*]] = nvvm.wmma.load %[[ADDRESS]], %[[LDM32]]
-    // CHECK32-SAME: {eltype = "f16", frag = "a", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}  : (!llvm.ptr<f16, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
+    // CHECK32-SAME: {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<a>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}  : (!llvm.ptr<f16, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK32:  llvm.return %[[FRAG]] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     return %0 : !gpu.mma_matrix<16x16xf16, "AOp">
   }
 }
 
 // -----
 
 gpu.module @test_module {
builtin.func @gpu_wmma_store_op(%arg0 : !gpu.mma_matrix<16x16xf16, "COp">) -> () {
     // CHECK:  %[[EL4:.*]] = llvm.extractvalue %[[D]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[BASE:.*]] = llvm.extractvalue %17[1] : !llvm.struct<(ptr<f16, 3>, ptr<f16, 3>, i64, array<2 x i64>, array<2 x i64>)>
     // CHECK:  %[[LDM:.*]] = llvm.mlir.constant(32 : index) : i64
     // CHECK:  %[[LI:.*]] = llvm.mul %[[INX]], %[[LDM]]   : i64
     // CHECK:  %[[LIJ:.*]] = llvm.add %[[LI]], %[[INX]]  : i64
     // CHECK:  %[[ADDRESS:.*]] = llvm.getelementptr %[[BASE]][%[[LIJ]]] : (!llvm.ptr<f16, 3>, i64) -> !llvm.ptr<f16, 3>
     // CHECK:  %[[LDM32:.*]] = llvm.mlir.constant(32 : index) : i32
     // CHECK:  nvvm.wmma.store %[[ADDRESS]], %[[LDM32]], %[[EL1]], %[[EL2]], %[[EL3]], %[[EL4]]
-    // CHECK-SAME: {eltype = "f16", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32} : !llvm.ptr<f16, 3>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>
+    // CHECK-SAME: {eltype = #nvvm.mma_type<f16>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : !llvm.ptr<f16, 3>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>
     // CHECK:  llvm.return
 
     // CHECK32:  %[[INX:.*]] = llvm.mlir.constant(16 : index) : i32
     // CHECK32: %{{.*}} = llvm.insertvalue %{{.*}}, %{{.*}}[{{.*}}, {{.*}}]
     // CHECK32:  %[[EL1:.*]] = llvm.extractvalue %[[D]][0 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK32:  %[[EL2:.*]] = llvm.extractvalue %[[D]][1 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK32:  %[[EL3:.*]] = llvm.extractvalue %[[D]][2 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK32:  %[[EL4:.*]] = llvm.extractvalue %[[D]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK32:  %[[BASE:.*]] = llvm.extractvalue %17[1] : !llvm.struct<(ptr<f16, 3>, ptr<f16, 3>, i32, array<2 x i32>, array<2 x i32>)>
     // CHECK32:  %[[LDM:.*]] = llvm.mlir.constant(32 : index) : i32
     // CHECK32:  %[[LI:.*]] = llvm.mul %[[INX]], %[[LDM]]   : i32
     // CHECK32:  %[[LIJ:.*]] = llvm.add %[[LI]], %[[INX]]  : i32
     // CHECK32:  %[[ADDRESS:.*]] = llvm.getelementptr %[[BASE]][%[[LIJ]]] : (!llvm.ptr<f16, 3>, i32) -> !llvm.ptr<f16, 3>
     // CHECK32:  %[[LDM32:.*]] = llvm.mlir.constant(32 : index) : i32
     // CHECK32:  nvvm.wmma.store %[[ADDRESS]], %[[LDM32]], %[[EL1]], %[[EL2]], %[[EL3]], %[[EL4]]
-    // CHECK32-SAME: {eltype = "f16", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32} : !llvm.ptr<f16, 3>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>
+    // CHECK32-SAME: {eltype = #nvvm.mma_type<f16>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : !llvm.ptr<f16, 3>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>
     // CHECK32:  llvm.return
     return
   }
 }
 
 // -----
 
 gpu.module @test_module {
builtin.func @gpu_wmma_mma_op(%A : !gpu.mma_matrix<16x16xf16, "AOp">, %B : !gpu.mma_matrix<16x16xf16, "BOp">, %C : !gpu.mma_matrix<16x16xf16, "COp">) -> (!gpu.mma_matrix<16x16xf16, "COp">) {
     // CHECK:  %[[B6:.*]] = llvm.extractvalue %[[B]][5 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[B7:.*]] = llvm.extractvalue %[[B]][6 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[B8:.*]] = llvm.extractvalue %[[B]][7 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[C1:.*]] = llvm.extractvalue %[[C]][0 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[C2:.*]] = llvm.extractvalue %[[C]][1 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[C3:.*]] = llvm.extractvalue %[[C]][2 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[C4:.*]] = llvm.extractvalue %[[C]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  %[[RES:.*]] = nvvm.wmma.mma %[[A1]], %[[A2]], %[[A3]], %[[A4]], %[[A5]], %[[A6]], %[[A7]], %[[A8]], %[[B1]], %[[B2]], %[[B3]], %[[B4]], %[[B5]], %[[B6]], %[[B7]], %[[B8]], %[[C1]], %[[C2]], %[[C3]], %[[C4]]
-    // CHECK-SAME: {eltypeA = "f16", eltypeB = "f16", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32} : (
+    // CHECK-SAME: {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f16>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : (
     // CHECK-SAME: vector<2xf16>, {{.*}}) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     // CHECK:  llvm.return %[[RES]] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
     return %D : !gpu.mma_matrix<16x16xf16, "COp">
   }
 }
 
 // -----
 
 gpu.module @test_module {
 
 // CHECK-LABEL: func @gpu_wmma_mma_loop_op
-//       CHECK:   %[[C:.+]] = nvvm.wmma.load %{{.*}}, %{{.*}} {eltype = "f16", frag = "c", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32} : (!llvm.ptr<f16>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
+//       CHECK:   %[[C:.+]] = nvvm.wmma.load %{{.*}}, %{{.*}} {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<c>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : (!llvm.ptr<f16>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   llvm.br ^bb1(%{{.*}}, %[[C]] : i64, !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>)
 //       CHECK:  ^bb1(%{{.*}}: i64, %[[ACC:.+]]: !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>):  // 2 preds: ^bb0, ^bb2
 //       CHECK:   llvm.cond_br %{{.*}}, ^bb2, ^bb3
 //       CHECK:  ^bb2:  // pred: ^bb1
-//       CHECK:   %[[A:.+]] = nvvm.wmma.load %{{.*}}, %{{.*}} {eltype = "f16", frag = "a", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32} : (!llvm.ptr<f16>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
-//       CHECK:   %[[B:.+]] = nvvm.wmma.load %{{.*}}, %{{.*}} {eltype = "f16", frag = "b", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32} : (!llvm.ptr<f16>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
+//       CHECK:   %[[A:.+]] = nvvm.wmma.load %{{.*}}, %{{.*}} {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<a>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : (!llvm.ptr<f16>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
+//       CHECK:   %[[B:.+]] = nvvm.wmma.load %{{.*}}, %{{.*}} {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<b>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : (!llvm.ptr<f16>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A0:.+]] = llvm.extractvalue %[[A]][0 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A1:.+]] = llvm.extractvalue %[[A]][1 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A2:.+]] = llvm.extractvalue %[[A]][2 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A3:.+]] = llvm.extractvalue %[[A]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A4:.+]] = llvm.extractvalue %[[A]][4 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A5:.+]] = llvm.extractvalue %[[A]][5 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A6:.+]] = llvm.extractvalue %[[A]][6 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[A7:.+]] = llvm.extractvalue %[[A]][7 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B0:.+]] = llvm.extractvalue %[[B]][0 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B1:.+]] = llvm.extractvalue %[[B]][1 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B2:.+]] = llvm.extractvalue %[[B]][2 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B3:.+]] = llvm.extractvalue %[[B]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B4:.+]] = llvm.extractvalue %[[B]][4 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B5:.+]] = llvm.extractvalue %[[B]][5 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B6:.+]] = llvm.extractvalue %[[B]][6 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[B7:.+]] = llvm.extractvalue %[[B]][7 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[ACC0:.+]] = llvm.extractvalue %[[ACC]][0 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[ACC1:.+]] = llvm.extractvalue %[[ACC]][1 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[ACC2:.+]] = llvm.extractvalue %[[ACC]][2 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[ACC3:.+]] = llvm.extractvalue %[[ACC]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
-//       CHECK:   %[[ACC_MUL:.+]] = nvvm.wmma.mma %[[A0]], %[[A1]], %[[A2]], %[[A3]], %[[A4]], %[[A5]], %[[A6]], %[[A7]], %[[B0]], %[[B1]], %[[B2]], %[[B3]], %[[B4]], %[[B5]], %[[B6]], %[[B7]], %[[ACC0]], %[[ACC1]], %[[ACC2]], %[[ACC3]] {eltypeA = "f16", eltypeB = "f16", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32} : (vector<2xf16>, {{.*}} -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
+//       CHECK:   %[[ACC_MUL:.+]] = nvvm.wmma.mma %[[A0]], %[[A1]], %[[A2]], %[[A3]], %[[A4]], %[[A5]], %[[A6]], %[[A7]], %[[B0]], %[[B1]], %[[B2]], %[[B3]], %[[B4]], %[[B5]], %[[B6]], %[[B7]], %[[ACC0]], %[[ACC1]], %[[ACC2]], %[[ACC3]] {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f16>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : (vector<2xf16>, {{.*}} -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   llvm.br ^bb1(%{{.*}}, %[[ACC_MUL]] : i64, !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>)
 //       CHECK:  ^bb3:  // pred: ^bb1
 //       CHECK:   %[[E0:.+]] = llvm.extractvalue %[[ACC]][0 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[E1:.+]] = llvm.extractvalue %[[ACC]][1 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[E2:.+]] = llvm.extractvalue %[[ACC]][2 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK:   %[[E3:.+]] = llvm.extractvalue %[[ACC]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
-//       CHECK:   nvvm.wmma.store %{{.*}}, %{{.*}}, %[[E0]], %[[E1]], %[[E2]], %[[E3]] {eltype = "f16", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32} : !llvm.ptr<f16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>
+//       CHECK:   nvvm.wmma.store %{{.*}}, %{{.*}}, %[[E0]], %[[E1]], %[[E2]], %[[E3]] {eltype = #nvvm.mma_type<f16>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32} : !llvm.ptr<f16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>
 
   builtin.func @gpu_wmma_mma_loop_op(%arg0: memref<128x128xf16>, %arg1: memref<128x128xf16>, %arg2: memref<128x128xf16>) {
       %c0 = arith.constant 0 : index
       %c128 = arith.constant 128 : index
       %c32 = arith.constant 32 : index
       %0 = gpu.subgroup_mma_load_matrix %arg2[%c0, %c0] {leadDimension = 128 : index} : memref<128x128xf16> -> !gpu.mma_matrix<16x16xf16, "COp">
       br ^bb1(%c0, %0 : index, !gpu.mma_matrix<16x16xf16, "COp">)
     ^bb1(%1: index, %2: !gpu.mma_matrix<16x16xf16, "COp">):  // 2 preds: ^bb0, ^bb2

 //       CHECK: %[[C2:.*]] = llvm.fadd %[[A2]], %[[B2]]  : vector<2xf16>
 //       CHECK: %[[M3:.*]] = llvm.insertvalue %[[C2]], %[[M2]][2 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK: %[[A3:.*]] = llvm.extractvalue %{{.*}}[3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK: %[[B3:.*]] = llvm.extractvalue %{{.*}}[3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK: %[[C3:.*]] = llvm.fadd %[[A3]], %[[B3]]  : vector<2xf16>
 //       CHECK: %[[M4:.*]] = llvm.insertvalue %[[C3]], %[[M3]][3 : i32] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 //       CHECK: llvm.return %[[M4]] : !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
   builtin.func @gpu_wmma_elementwise(%A : !gpu.mma_matrix<16x16xf16, "COp">, %B : !gpu.mma_matrix<16x16xf16, "COp">)  ->(!gpu.mma_matrix<16x16xf16, "COp">) {
-    %C = gpu.subgroup_mma_elementwise %A, %B { operation = "ADDF" } :
-      (!gpu.mma_matrix<16x16xf16, "COp">, !gpu.mma_matrix<16x16xf16, "COp">) -> !gpu.mma_matrix<16x16xf16, "COp">
+    %C = gpu.subgroup_mma_elementwise addf %A, %B : (!gpu.mma_matrix<16x16xf16, "COp">, !gpu.mma_matrix<16x16xf16, "COp">) -> !gpu.mma_matrix<16x16xf16, "COp">
     return %C : !gpu.mma_matrix<16x16xf16, "COp">
   }
 }

mlir/test/Conversion/GPUToROCDL/gpu-to-rocdl.mlir

 // RUN: mlir-opt %s -convert-gpu-to-rocdl -split-input-file | FileCheck %s
 // RUN: mlir-opt %s -convert-gpu-to-rocdl='index-bitwidth=32' -split-input-file | FileCheck --check-prefix=CHECK32 %s
 
 gpu.module @test_module {
   // CHECK-LABEL: func @gpu_index_ops()
   // CHECK32-LABEL: func @gpu_index_ops()
   builtin.func @gpu_index_ops()
       -> (index, index, index, index, index, index,
           index, index, index, index, index, index) {
     // CHECK32-NOT: = llvm.sext %{{.*}} : i32 to i64
 
     // CHECK: rocdl.workitem.id.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %tIdX = "gpu.thread_id"() {dimension = "x"} : () -> (index)
+    %tIdX = gpu.thread_id x
     // CHECK: rocdl.workitem.id.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %tIdY = "gpu.thread_id"() {dimension = "y"} : () -> (index)
+    %tIdY = gpu.thread_id y
     // CHECK: rocdl.workitem.id.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %tIdZ = "gpu.thread_id"() {dimension = "z"} : () -> (index)
+    %tIdZ = gpu.thread_id z
 
     // CHECK: rocdl.workgroup.dim.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bDimX = "gpu.block_dim"() {dimension = "x"} : () -> (index)
+    %bDimX = gpu.block_dim x
     // CHECK: rocdl.workgroup.dim.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bDimY = "gpu.block_dim"() {dimension = "y"} : () -> (index)
+    %bDimY = gpu.block_dim y
     // CHECK: rocdl.workgroup.dim.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bDimZ = "gpu.block_dim"() {dimension = "z"} : () -> (index)
+    %bDimZ = gpu.block_dim z
 
     // CHECK: rocdl.workgroup.id.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bIdX = "gpu.block_id"() {dimension = "x"} : () -> (index)
+    %bIdX = gpu.block_id x
     // CHECK: rocdl.workgroup.id.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bIdY = "gpu.block_id"() {dimension = "y"} : () -> (index)
+    %bIdY = gpu.block_id y
     // CHECK: rocdl.workgroup.id.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %bIdZ = "gpu.block_id"() {dimension = "z"} : () -> (index)
+    %bIdZ = gpu.block_id z
 
     // CHECK: rocdl.grid.dim.x : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %gDimX = "gpu.grid_dim"() {dimension = "x"} : () -> (index)
+    %gDimX = gpu.grid_dim x
     // CHECK: rocdl.grid.dim.y : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %gDimY = "gpu.grid_dim"() {dimension = "y"} : () -> (index)
+    %gDimY = gpu.grid_dim y
     // CHECK: rocdl.grid.dim.z : i32
     // CHECK: = llvm.sext %{{.*}} : i32 to i64
-    %gDimZ = "gpu.grid_dim"() {dimension = "z"} : () -> (index)
+    %gDimZ = gpu.grid_dim z
 
     std.return %tIdX, %tIdY, %tIdZ, %bDimX, %bDimY, %bDimZ,
                %bIdX, %bIdY, %bIdZ, %gDimX, %gDimY, %gDimZ
         : index, index, index, index, index, index,
           index, index, index, index, index, index
   }
 }
 

mlir/test/Conversion/GPUToSPIRV/builtins.mlir

module attributes {gpu.container_module} {
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   // CHECK: spv.GlobalVariable [[WORKGROUPID:@.*]] built_in("WorkgroupId")
   gpu.module @kernels {
     gpu.func @builtin_workgroup_id_x() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[16, 1, 1]>: vector<3xi32>}} {
       // CHECK: [[ADDRESS:%.*]] = spv.mlir.addressof [[WORKGROUPID]]
       // CHECK-NEXT: [[VEC:%.*]] = spv.Load "Input" [[ADDRESS]]
       // CHECK-NEXT: {{%.*}} = spv.CompositeExtract [[VEC]]{{\[}}0 : i32{{\]}}
-      %0 = "gpu.block_id"() {dimension = "x"} : () -> index
+      %0 = gpu.block_id x
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {
module attributes {gpu.container_module} {
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   // CHECK: spv.GlobalVariable [[WORKGROUPID:@.*]] built_in("WorkgroupId")
   gpu.module @kernels {
     gpu.func @builtin_workgroup_id_y() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[16, 1, 1]>: vector<3xi32>}} {
       // CHECK: [[ADDRESS:%.*]] = spv.mlir.addressof [[WORKGROUPID]]
       // CHECK-NEXT: [[VEC:%.*]] = spv.Load "Input" [[ADDRESS]]
       // CHECK-NEXT: {{%.*}} = spv.CompositeExtract [[VEC]]{{\[}}1 : i32{{\]}}
-      %0 = "gpu.block_id"() {dimension = "y"} : () -> index
+      %0 = gpu.block_id y
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {
   func @builtin() {
     %c0 = arith.constant 1 : index
     gpu.launch_func @kernels::@builtin_workgroup_id_z
         blocks in (%c0, %c0, %c0) threads in (%c0, %c0, %c0)
     return
   }
 
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   // CHECK: spv.GlobalVariable [[WORKGROUPID:@.*]] built_in("WorkgroupId")
   gpu.module @kernels {
     gpu.func @builtin_workgroup_id_z() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[16, 1, 1]>: vector<3xi32>}} {
       // CHECK: [[ADDRESS:%.*]] = spv.mlir.addressof [[WORKGROUPID]]
       // CHECK-NEXT: [[VEC:%.*]] = spv.Load "Input" [[ADDRESS]]
       // CHECK-NEXT: {{%.*}} = spv.CompositeExtract [[VEC]]{{\[}}2 : i32{{\]}}
-      %0 = "gpu.block_id"() {dimension = "z"} : () -> index
+      %0 = gpu.block_id z
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {
   func @builtin() {
     %c0 = arith.constant 1 : index
     gpu.launch_func @kernels::@builtin_workgroup_size_x
         blocks in (%c0, %c0, %c0) threads in (%c0, %c0, %c0)
     return
   }
 
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   gpu.module @kernels {
     gpu.func @builtin_workgroup_size_x() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[32, 1, 1]>: vector<3xi32>}} {
       // The constant value is obtained from the spv.entry_point_abi.
       // Note that this ignores the workgroup size specification in gpu.launch.
       // We may want to define gpu.workgroup_size and convert it to the entry
       // point ABI we want here.
       // CHECK: spv.Constant 32 : i32
-      %0 = "gpu.block_dim"() {dimension = "x"} : () -> index
+      %0 = gpu.block_dim x
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {
   func @builtin() {
     %c0 = arith.constant 1 : index
     gpu.launch_func @kernels::@builtin_workgroup_size_y
         blocks in (%c0, %c0, %c0) threads in (%c0, %c0, %c0)
     return
   }
 
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   gpu.module @kernels {
     gpu.func @builtin_workgroup_size_y() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[32, 4, 1]>: vector<3xi32>}} {
       // The constant value is obtained from the spv.entry_point_abi.
       // CHECK: spv.Constant 4 : i32
-      %0 = "gpu.block_dim"() {dimension = "y"} : () -> index
+      %0 = gpu.block_dim y
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {
   func @builtin() {
     %c0 = arith.constant 1 : index
     gpu.launch_func @kernels::@builtin_workgroup_size_z
         blocks in (%c0, %c0, %c0) threads in (%c0, %c0, %c0)
     return
   }
 
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   gpu.module @kernels {
     gpu.func @builtin_workgroup_size_z() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[32, 4, 1]>: vector<3xi32>}} {
       // The constant value is obtained from the spv.entry_point_abi.
       // CHECK: spv.Constant 1 : i32
-      %0 = "gpu.block_dim"() {dimension = "z"} : () -> index
+      %0 = gpu.block_dim z
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {
   func @builtin() {
     %c0 = arith.constant 1 : index
     gpu.launch_func @kernels::@builtin_local_id_x
         blocks in (%c0, %c0, %c0) threads in (%c0, %c0, %c0)
     return
   }
 
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   // CHECK: spv.GlobalVariable [[LOCALINVOCATIONID:@.*]] built_in("LocalInvocationId")
   gpu.module @kernels {
     gpu.func @builtin_local_id_x() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[16, 1, 1]>: vector<3xi32>}} {
       // CHECK: [[ADDRESS:%.*]] = spv.mlir.addressof [[LOCALINVOCATIONID]]
       // CHECK-NEXT: [[VEC:%.*]] = spv.Load "Input" [[ADDRESS]]
       // CHECK-NEXT: {{%.*}} = spv.CompositeExtract [[VEC]]{{\[}}0 : i32{{\]}}
-      %0 = "gpu.thread_id"() {dimension = "x"} : () -> index
+      %0 = gpu.thread_id x
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {
   func @builtin() {
     %c0 = arith.constant 1 : index
     gpu.launch_func @kernels::@builtin_num_workgroups_x
         blocks in (%c0, %c0, %c0) threads in (%c0, %c0, %c0)
     return
   }
 
   // CHECK-LABEL:  spv.module @{{.*}} Logical GLSL450
   // CHECK: spv.GlobalVariable [[NUMWORKGROUPS:@.*]] built_in("NumWorkgroups")
   gpu.module @kernels {
     gpu.func @builtin_num_workgroups_x() kernel
       attributes {spv.entry_point_abi = {local_size = dense<[16, 1, 1]>: vector<3xi32>}} {
       // CHECK: [[ADDRESS:%.*]] = spv.mlir.addressof [[NUMWORKGROUPS]]
       // CHECK-NEXT: [[VEC:%.*]] = spv.Load "Input" [[ADDRESS]]
       // CHECK-NEXT: {{%.*}} = spv.CompositeExtract [[VEC]]{{\[}}0 : i32{{\]}}
-      %0 = "gpu.grid_dim"() {dimension = "x"} : () -> index
+      %0 = gpu.grid_dim x
       gpu.return
     }
   }
 }
 
 // -----
 
 module attributes {gpu.container_module} {

mlir/test/Conversion/GPUToSPIRV/load-store.mlir

gpu.module @kernels {
     gpu.func @load_store_kernel(%arg0: memref<12x4xf32>, %arg1: memref<12x4xf32>, %arg2: memref<12x4xf32>, %arg3: index, %arg4: index, %arg5: index, %arg6: index) kernel
       attributes {spv.entry_point_abi = {local_size = dense<[16, 1, 1]>: vector<3xi32>}} {
       // CHECK: %[[ADDRESSWORKGROUPID:.*]] = spv.mlir.addressof @[[$WORKGROUPIDVAR]]
       // CHECK: %[[WORKGROUPID:.*]] = spv.Load "Input" %[[ADDRESSWORKGROUPID]]
       // CHECK: %[[WORKGROUPIDX:.*]] = spv.CompositeExtract %[[WORKGROUPID]]{{\[}}0 : i32{{\]}}
       // CHECK: %[[ADDRESSLOCALINVOCATIONID:.*]] = spv.mlir.addressof @[[$LOCALINVOCATIONIDVAR]]
       // CHECK: %[[LOCALINVOCATIONID:.*]] = spv.Load "Input" %[[ADDRESSLOCALINVOCATIONID]]
       // CHECK: %[[LOCALINVOCATIONIDX:.*]] = spv.CompositeExtract %[[LOCALINVOCATIONID]]{{\[}}0 : i32{{\]}}
-      %0 = "gpu.block_id"() {dimension = "x"} : () -> index
-      %1 = "gpu.block_id"() {dimension = "y"} : () -> index
-      %2 = "gpu.block_id"() {dimension = "z"} : () -> index
-      %3 = "gpu.thread_id"() {dimension = "x"} : () -> index
-      %4 = "gpu.thread_id"() {dimension = "y"} : () -> index
-      %5 = "gpu.thread_id"() {dimension = "z"} : () -> index
-      %6 = "gpu.grid_dim"() {dimension = "x"} : () -> index
-      %7 = "gpu.grid_dim"() {dimension = "y"} : () -> index
-      %8 = "gpu.grid_dim"() {dimension = "z"} : () -> index
-      %9 = "gpu.block_dim"() {dimension = "x"} : () -> index
-      %10 = "gpu.block_dim"() {dimension = "y"} : () -> index
-      %11 = "gpu.block_dim"() {dimension = "z"} : () -> index
+      %0 = gpu.block_id x
+      %1 = gpu.block_id y
+      %2 = gpu.block_id z
+      %3 = gpu.thread_id x
+      %4 = gpu.thread_id y
+      %5 = gpu.thread_id z
+      %6 = gpu.grid_dim x
+      %7 = gpu.grid_dim y
+      %8 = gpu.grid_dim z
+      %9 = gpu.block_dim x
+      %10 = gpu.block_dim y
+      %11 = gpu.block_dim z
       // CHECK: %[[INDEX1:.*]] = spv.IAdd %[[ARG3]], %[[WORKGROUPIDX]]
       %12 = arith.addi %arg3, %0 : index
       // CHECK: %[[INDEX2:.*]] = spv.IAdd %[[ARG4]], %[[LOCALINVOCATIONIDX]]
       %13 = arith.addi %arg4, %3 : index
       // CHECK: %[[ZERO:.*]] = spv.Constant 0 : i32
       // CHECK: %[[OFFSET1_0:.*]] = spv.Constant 0 : i32
       // CHECK: %[[STRIDE1_1:.*]] = spv.Constant 4 : i32
       // CHECK: %[[UPDATE1_1:.*]] = spv.IMul %[[STRIDE1_1]], %[[INDEX1]] : i32

mlir/test/Conversion/VectorToGPU/vector-to-mma-ops.mlir

 // CHECK-LABEL: func @matmul_fused_elementwise
 //   CHECK-DAG:   %[[CST_0:.+]] = arith.constant 0.000000e+00 : f16
 //   CHECK-DAG:   %[[CST_1:.+]] = arith.constant 1.000000e+00 : f16
 //   CHECK-DAG:   %[[A:.+]] = gpu.subgroup_mma_load_matrix %{{.*}}[%{{.*}}, %{{.*}}] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "AOp">
 //   CHECK-DAG:   %[[B:.+]] = gpu.subgroup_mma_load_matrix %{{.*}}[%c0, %c0] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "BOp">
 //   CHECK-DAG:   %[[C0:.+]] = gpu.subgroup_mma_constant_matrix %[[CST_0]] : !gpu.mma_matrix<16x16xf16, "COp">
 //   CHECK-DAG:   %[[C1:.+]] = gpu.subgroup_mma_constant_matrix %[[CST_1]] : !gpu.mma_matrix<16x16xf16, "COp">
 //       CHECK:   %[[D:.+]] = gpu.subgroup_mma_compute %[[A]], %[[B]], %[[C0]] : !gpu.mma_matrix<16x16xf16, "AOp">, !gpu.mma_matrix<16x16xf16, "BOp"> -> !gpu.mma_matrix<16x16xf16, "COp">
-//       CHECK:   %[[E:.+]] = gpu.subgroup_mma_elementwise %[[D]], %[[C1]] {operation = "ADDF"} : (!gpu.mma_matrix<16x16xf16, "COp">, !gpu.mma_matrix<16x16xf16, "COp">) -> !gpu.mma_matrix<16x16xf16, "COp">
+//       CHECK:   %[[E:.+]] = gpu.subgroup_mma_elementwise addf %[[D]], %[[C1]] : (!gpu.mma_matrix<16x16xf16, "COp">, !gpu.mma_matrix<16x16xf16, "COp">) -> !gpu.mma_matrix<16x16xf16, "COp">
 //       CHECK:   gpu.subgroup_mma_store_matrix %[[E]], %{{.*}}[%{{.*}}, %{{.*}}] {leadDimension = 16 : index} : !gpu.mma_matrix<16x16xf16, "COp">, memref<16x16xf16>
 func @matmul_fused_elementwise(%arg0: memref<16x16xf16>, %arg1: memref<16x16xf16>, %arg2: memref<16x16xf16>) {
   %cst_0 = arith.constant dense<0.000000e+00> : vector<16x16xf16>
   %cst_1 = arith.constant dense<1.000000e+00> : vector<16x16xf16>
   %c0 = arith.constant 0 : index
   %cst = arith.constant 0.000000e+00 : f16
   %A = vector.transfer_read %arg0[%c0, %c0], %cst {in_bounds = [true, true]} : memref<16x16xf16>, vector<16x16xf16>
   %B = vector.transfer_read %arg1[%c0, %c0], %cst {permutation_map = #map0, in_bounds = [true, true]} : memref<16x16xf16>, vector<16x16xf16>
   %D = vector.contract {indexing_maps = [#map1, #map2, #map3], iterator_types = ["parallel", "parallel", "reduction"], kind = #vector.kind<add>} %A, %B, %cst_0 : vector<16x16xf16>, vector<16x16xf16> into vector<16x16xf16>
   %E = arith.addf %D, %cst_1 : vector<16x16xf16>
   vector.transfer_write %E, %arg2[%c0, %c0] {in_bounds = [true, true]} : vector<16x16xf16>, memref<16x16xf16>
   return
 }
 
 // CHECK-LABEL: func @matmul_fused_broadcast
 //   CHECK-DAG:   %[[CST_0:.+]] = arith.constant 0.000000e+00 : f16
 //   CHECK-DAG:   %[[A:.+]] = gpu.subgroup_mma_load_matrix %{{.*}}[%{{.*}}, %{{.*}}] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "AOp">
 //   CHECK-DAG:   %[[B:.+]] = gpu.subgroup_mma_load_matrix %{{.*}}[%{{.*}}, %{{.*}}] {leadDimension = 16 : index} : memref<16x16xf16> -> !gpu.mma_matrix<16x16xf16, "BOp">
 //   CHECK-DAG:   %[[C0:.+]] = gpu.subgroup_mma_constant_matrix %[[CST_0]] : !gpu.mma_matrix<16x16xf16, "COp">
 //       CHECK:   %[[D:.+]] = gpu.subgroup_mma_compute %[[A]], %[[B]], %[[C0]] : !gpu.mma_matrix<16x16xf16, "AOp">, !gpu.mma_matrix<16x16xf16, "BOp"> -> !gpu.mma_matrix<16x16xf16, "COp">
 //       CHECK:   %[[E:.+]] = gpu.subgroup_mma_load_matrix %{{.*}}[%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}}] {leadDimension = 0 : index} : memref<16x16x16x16xf16> -> !gpu.mma_matrix<16x16xf16, "COp">
-//       CHECK:   %[[F:.+]] = gpu.subgroup_mma_elementwise %[[D]], %[[E]] {operation = "DIVF"} : (!gpu.mma_matrix<16x16xf16, "COp">, !gpu.mma_matrix<16x16xf16, "COp">) -> !gpu.mma_matrix<16x16xf16, "COp">
+//       CHECK:   %[[F:.+]] = gpu.subgroup_mma_elementwise divf %[[D]], %[[E]] : (!gpu.mma_matrix<16x16xf16, "COp">, !gpu.mma_matrix<16x16xf16, "COp">) -> !gpu.mma_matrix<16x16xf16, "COp">
 //       CHECK:   gpu.subgroup_mma_store_matrix %[[F]], %{{.*}}[%{{.*}}, %{{.*}}] {leadDimension = 16 : index} : !gpu.mma_matrix<16x16xf16, "COp">, memref<16x16xf16>
 func @matmul_fused_broadcast(%arg0: memref<16x16xf16>, %arg1: memref<16x16xf16>,
   %arg2: memref<16x16xf16>, %arg3: memref<16x16x16x16xf16>) {
   %cst_0 = arith.constant dense<0.000000e+00> : vector<16x16xf16>
   %c0 = arith.constant 0 : index
   %cst = arith.constant 0.000000e+00 : f16
   %A = vector.transfer_read %arg0[%c0, %c0], %cst {in_bounds = [true, true]} : memref<16x16xf16>, vector<16x16xf16>
   %B = vector.transfer_read %arg1[%c0, %c0], %cst {permutation_map = #map0, in_bounds = [true, true]} : memref<16x16xf16>, vector<16x16xf16>
   %D = vector.contract {indexing_maps = [#map1, #map2, #map3], iterator_types = ["parallel", "parallel", "reduction"], kind = #vector.kind<add>} %A, %B, %cst_0 : vector<16x16xf16>, vector<16x16xf16> into vector<16x16xf16>
   %E = vector.transfer_read %arg3[%c0, %c0, %c0, %c0], %cst
     {in_bounds = [true, true], permutation_map = affine_map<(d0, d1, d2, d3)->(0, d3)>}
     : memref<16x16x16x16xf16>, vector<16x16xf16>
   %F = arith.divf %D, %E : vector<16x16xf16>
   vector.transfer_write %F, %arg2[%c0, %c0] {in_bounds = [true, true]} : vector<16x16xf16>, memref<16x16xf16>
   return
 }

mlir/test/Dialect/GPU/all-reduce-max.mlir

gpu.func @kernel(%arg0 : f32) kernel {
     // CHECK-DAG:   [[VAL_3:%.*]] = arith.constant 0 : i32
     // CHECK-DAG:   [[VAL_4:%.*]] = arith.constant 0 : index
     // CHECK-DAG:   [[VAL_5:%.*]] = arith.constant 32 : i32
     // CHECK-DAG:   [[VAL_6:%.*]] = arith.constant 1 : i32
     // CHECK-DAG:   [[VAL_7:%.*]] = arith.constant 2 : i32
     // CHECK-DAG:   [[VAL_8:%.*]] = arith.constant 4 : i32
     // CHECK-DAG:   [[VAL_9:%.*]] = arith.constant 8 : i32
     // CHECK-DAG:   [[VAL_10:%.*]] = arith.constant 16 : i32
-    // CHECK:   [[VAL_11:%.*]] = "gpu.block_dim"() {dimension = "x"} : () -> index
+    // CHECK:   [[VAL_11:%.*]] = gpu.block_dim x
     // CHECK:   [[VAL_12:%.*]] = arith.index_cast [[VAL_11]] : index to i32
-    // CHECK:   [[VAL_13:%.*]] = "gpu.block_dim"() {dimension = "y"} : () -> index
+    // CHECK:   [[VAL_13:%.*]] = gpu.block_dim y
     // CHECK:   [[VAL_14:%.*]] = arith.index_cast [[VAL_13]] : index to i32
-    // CHECK:   [[VAL_15:%.*]] = "gpu.block_dim"() {dimension = "z"} : () -> index
+    // CHECK:   [[VAL_15:%.*]] = gpu.block_dim z
     // CHECK:   [[VAL_16:%.*]] = arith.index_cast [[VAL_15]] : index to i32
-    // CHECK:   [[VAL_17:%.*]] = "gpu.thread_id"() {dimension = "x"} : () -> index
+    // CHECK:   [[VAL_17:%.*]] = gpu.thread_id x
     // CHECK:   [[VAL_18:%.*]] = arith.index_cast [[VAL_17]] : index to i32
-    // CHECK:   [[VAL_19:%.*]] = "gpu.thread_id"() {dimension = "y"} : () -> index
+    // CHECK:   [[VAL_19:%.*]] = gpu.thread_id y
     // CHECK:   [[VAL_20:%.*]] = arith.index_cast [[VAL_19]] : index to i32
-    // CHECK:   [[VAL_21:%.*]] = "gpu.thread_id"() {dimension = "z"} : () -> index
+    // CHECK:   [[VAL_21:%.*]] = gpu.thread_id z
     // CHECK:   [[VAL_22:%.*]] = arith.index_cast [[VAL_21]] : index to i32
     // CHECK:   [[VAL_23:%.*]] = arith.muli [[VAL_22]], [[VAL_14]] : i32
     // CHECK:   [[VAL_24:%.*]] = arith.addi [[VAL_23]], [[VAL_20]] : i32
     // CHECK:   [[VAL_25:%.*]] = arith.muli [[VAL_24]], [[VAL_12]] : i32
     // CHECK:   [[VAL_26:%.*]] = arith.muli [[VAL_12]], [[VAL_14]] : i32
     // CHECK:   [[VAL_27:%.*]] = arith.addi [[VAL_25]], [[VAL_18]] : i32
     // CHECK:   [[VAL_28:%.*]] = arith.muli [[VAL_26]], [[VAL_16]] : i32
     // CHECK:   [[VAL_29:%.*]] = arith.andi [[VAL_27]], [[VAL_2]] : i32
     // CHECK:   [[VAL_30:%.*]] = arith.cmpi eq, [[VAL_29]], [[VAL_3]] : i32
     // CHECK:   [[VAL_31:%.*]] = arith.subi [[VAL_27]], [[VAL_29]] : i32
     // CHECK:   [[VAL_32:%.*]] = arith.subi [[VAL_28]], [[VAL_31]] : i32
     // CHECK:   [[VAL_33:%.*]] = arith.cmpi slt, [[VAL_32]], [[VAL_5]] : i32
     // CHECK:   cond_br [[VAL_33]], ^bb1, ^bb17
     // CHECK: ^bb1:
-    // CHECK:   [[VAL_34:%.*]], [[VAL_35:%.*]] = gpu.shuffle [[VAL_0]], [[VAL_6]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_34:%.*]], [[VAL_35:%.*]] = gpu.shuffle xor [[VAL_0]], [[VAL_6]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_35]], ^bb2, ^bb3
     // CHECK: ^bb2:
     // CHECK:   [[VAL_36:%.*]] = arith.cmpf ugt, [[VAL_0]], [[VAL_34]] : f32
     // CHECK:   [[VAL_37:%.*]] = select [[VAL_36]], [[VAL_0]], [[VAL_34]] : f32
     // CHECK:   br ^bb4([[VAL_37]] : f32)
     // CHECK: ^bb3:
     // CHECK:   br ^bb4([[VAL_0]] : f32)
     // CHECK: ^bb4([[VAL_38:%.*]]: f32):
-    // CHECK:   [[VAL_39:%.*]], [[VAL_40:%.*]] = gpu.shuffle [[VAL_38]], [[VAL_7]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_39:%.*]], [[VAL_40:%.*]] = gpu.shuffle xor [[VAL_38]], [[VAL_7]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_40]], ^bb5, ^bb6
     // CHECK: ^bb5:
     // CHECK:   [[VAL_41:%.*]] = arith.cmpf ugt, [[VAL_38]], [[VAL_39]] : f32
     // CHECK:   [[VAL_42:%.*]] = select [[VAL_41]], [[VAL_38]], [[VAL_39]] : f32
     // CHECK:   br ^bb7([[VAL_42]] : f32)
     // CHECK: ^bb6:
     // CHECK:   br ^bb7([[VAL_38]] : f32)
     // CHECK: ^bb7([[VAL_43:%.*]]: f32):
-    // CHECK:   [[VAL_44:%.*]], [[VAL_45:%.*]] = gpu.shuffle [[VAL_43]], [[VAL_8]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_44:%.*]], [[VAL_45:%.*]] = gpu.shuffle xor [[VAL_43]], [[VAL_8]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_45]], ^bb8, ^bb9
     // CHECK: ^bb8:
     // CHECK:   [[VAL_46:%.*]] = arith.cmpf ugt, [[VAL_43]], [[VAL_44]] : f32
     // CHECK:   [[VAL_47:%.*]] = select [[VAL_46]], [[VAL_43]], [[VAL_44]] : f32
     // CHECK:   br ^bb10([[VAL_47]] : f32)
     // CHECK: ^bb9:
     // CHECK:   br ^bb10([[VAL_43]] : f32)
     // CHECK: ^bb10([[VAL_48:%.*]]: f32):
-    // CHECK:   [[VAL_49:%.*]], [[VAL_50:%.*]] = gpu.shuffle [[VAL_48]], [[VAL_9]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_49:%.*]], [[VAL_50:%.*]] = gpu.shuffle xor [[VAL_48]], [[VAL_9]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_50]], ^bb11, ^bb12
     // CHECK: ^bb11:
     // CHECK:   [[VAL_51:%.*]] = arith.cmpf ugt, [[VAL_48]], [[VAL_49]] : f32
     // CHECK:   [[VAL_52:%.*]] = select [[VAL_51]], [[VAL_48]], [[VAL_49]] : f32
     // CHECK:   br ^bb13([[VAL_52]] : f32)
     // CHECK: ^bb12:
     // CHECK:   br ^bb13([[VAL_48]] : f32)
     // CHECK: ^bb13([[VAL_53:%.*]]: f32):
-    // CHECK:   [[VAL_54:%.*]], [[VAL_55:%.*]] = gpu.shuffle [[VAL_53]], [[VAL_10]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_54:%.*]], [[VAL_55:%.*]] = gpu.shuffle xor [[VAL_53]], [[VAL_10]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_55]], ^bb14, ^bb15
     // CHECK: ^bb14:
     // CHECK:   [[VAL_56:%.*]] = arith.cmpf ugt, [[VAL_53]], [[VAL_54]] : f32
     // CHECK:   [[VAL_57:%.*]] = select [[VAL_56]], [[VAL_53]], [[VAL_54]] : f32
     // CHECK:   br ^bb16([[VAL_57]] : f32)
     // CHECK: ^bb15:
     // CHECK:   br ^bb16([[VAL_53]] : f32)
     // CHECK: ^bb16([[VAL_58:%.*]]: f32):
     // CHECK:   br ^bb18([[VAL_58]] : f32)
     // CHECK: ^bb17:
-    // CHECK:   [[VAL_59:%.*]], [[VAL_60:%.*]] = gpu.shuffle [[VAL_0]], [[VAL_6]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_59:%.*]], [[VAL_60:%.*]] = gpu.shuffle xor [[VAL_0]], [[VAL_6]], [[VAL_5]] : f32
     // CHECK:   [[VAL_61:%.*]] = arith.cmpf ugt, [[VAL_0]], [[VAL_59]] : f32
     // CHECK:   [[VAL_62:%.*]] = select [[VAL_61]], [[VAL_0]], [[VAL_59]] : f32
-    // CHECK:   [[VAL_63:%.*]], [[VAL_64:%.*]] = gpu.shuffle [[VAL_62]], [[VAL_7]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_63:%.*]], [[VAL_64:%.*]] = gpu.shuffle xor [[VAL_62]], [[VAL_7]], [[VAL_5]] : f32
     // CHECK:   [[VAL_65:%.*]] = arith.cmpf ugt, [[VAL_62]], [[VAL_63]] : f32
     // CHECK:   [[VAL_66:%.*]] = select [[VAL_65]], [[VAL_62]], [[VAL_63]] : f32
-    // CHECK:   [[VAL_67:%.*]], [[VAL_68:%.*]] = gpu.shuffle [[VAL_66]], [[VAL_8]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_67:%.*]], [[VAL_68:%.*]] = gpu.shuffle xor [[VAL_66]], [[VAL_8]], [[VAL_5]] : f32
     // CHECK:   [[VAL_69:%.*]] = arith.cmpf ugt, [[VAL_66]], [[VAL_67]] : f32
     // CHECK:   [[VAL_70:%.*]] = select [[VAL_69]], [[VAL_66]], [[VAL_67]] : f32
-    // CHECK:   [[VAL_71:%.*]], [[VAL_72:%.*]] = gpu.shuffle [[VAL_70]], [[VAL_9]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_71:%.*]], [[VAL_72:%.*]] = gpu.shuffle xor [[VAL_70]], [[VAL_9]], [[VAL_5]] : f32
     // CHECK:   [[VAL_73:%.*]] = arith.cmpf ugt, [[VAL_70]], [[VAL_71]] : f32
     // CHECK:   [[VAL_74:%.*]] = select [[VAL_73]], [[VAL_70]], [[VAL_71]] : f32
-    // CHECK:   [[VAL_75:%.*]], [[VAL_76:%.*]] = gpu.shuffle [[VAL_74]], [[VAL_10]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_75:%.*]], [[VAL_76:%.*]] = gpu.shuffle xor [[VAL_74]], [[VAL_10]], [[VAL_5]] : f32
     // CHECK:   [[VAL_77:%.*]] = arith.cmpf ugt, [[VAL_74]], [[VAL_75]] : f32
     // CHECK:   [[VAL_78:%.*]] = select [[VAL_77]], [[VAL_74]], [[VAL_75]] : f32
     // CHECK:   br ^bb18([[VAL_78]] : f32)
     // CHECK: ^bb18([[VAL_79:%.*]]: f32):
     // CHECK:   cond_br [[VAL_30]], ^bb19, ^bb20
     // CHECK: ^bb19:
     // CHECK:   [[VAL_80:%.*]] = arith.divsi [[VAL_27]], [[VAL_5]] : i32
     // CHECK:   [[VAL_81:%.*]] = arith.index_cast [[VAL_80]] : i32 to index
     // CHECK:   store [[VAL_79]], [[VAL_1]]{{\[}}[[VAL_81]]] : memref<32xf32, 3>
     // CHECK:   br ^bb21
     // CHECK: ^bb20:
     // CHECK:   br ^bb21
     // CHECK: ^bb21:
     // CHECK:   gpu.barrier
     // CHECK:   [[VAL_82:%.*]] = arith.addi [[VAL_28]], [[VAL_2]] : i32
     // CHECK:   [[VAL_83:%.*]] = arith.divsi [[VAL_82]], [[VAL_5]] : i32
     // CHECK:   [[VAL_84:%.*]] = arith.cmpi slt, [[VAL_27]], [[VAL_83]] : i32
     // CHECK:   cond_br [[VAL_84]], ^bb22, ^bb41
     // CHECK: ^bb22:
     // CHECK:   [[VAL_85:%.*]] = arith.index_cast [[VAL_27]] : i32 to index
     // CHECK:   [[VAL_86:%.*]] = memref.load [[VAL_1]]{{\[}}[[VAL_85]]] : memref<32xf32, 3>
     // CHECK:   [[VAL_87:%.*]] = arith.cmpi slt, [[VAL_83]], [[VAL_5]] : i32
     // CHECK:   cond_br [[VAL_87]], ^bb23, ^bb39
     // CHECK: ^bb23:
-    // CHECK:   [[VAL_88:%.*]], [[VAL_89:%.*]] = gpu.shuffle [[VAL_86]], [[VAL_6]], [[VAL_83]] xor : f32
+    // CHECK:   [[VAL_88:%.*]], [[VAL_89:%.*]] = gpu.shuffle xor [[VAL_86]], [[VAL_6]], [[VAL_83]] : f32
     // CHECK:   cond_br [[VAL_89]], ^bb24, ^bb25
     // CHECK: ^bb24:
     // CHECK:   [[VAL_90:%.*]] = arith.cmpf ugt, [[VAL_86]], [[VAL_88]] : f32
     // CHECK:   [[VAL_91:%.*]] = select [[VAL_90]], [[VAL_86]], [[VAL_88]] : f32
     // CHECK:   br ^bb26([[VAL_91]] : f32)
     // CHECK: ^bb25:
     // CHECK:   br ^bb26([[VAL_86]] : f32)
     // CHECK: ^bb26([[VAL_92:%.*]]: f32):
-    // CHECK:   [[VAL_93:%.*]], [[VAL_94:%.*]] = gpu.shuffle [[VAL_92]], [[VAL_7]], [[VAL_83]] xor : f32
+    // CHECK:   [[VAL_93:%.*]], [[VAL_94:%.*]] = gpu.shuffle xor [[VAL_92]], [[VAL_7]], [[VAL_83]] : f32
     // CHECK:   cond_br [[VAL_94]], ^bb27, ^bb28
     // CHECK: ^bb27:
     // CHECK:   [[VAL_95:%.*]] = arith.cmpf ugt, [[VAL_92]], [[VAL_93]] : f32
     // CHECK:   [[VAL_96:%.*]] = select [[VAL_95]], [[VAL_92]], [[VAL_93]] : f32
     // CHECK:   br ^bb29([[VAL_96]] : f32)
     // CHECK: ^bb28:
     // CHECK:   br ^bb29([[VAL_92]] : f32)
     // CHECK: ^bb29([[VAL_97:%.*]]: f32):
-    // CHECK:   [[VAL_98:%.*]], [[VAL_99:%.*]] = gpu.shuffle [[VAL_97]], [[VAL_8]], [[VAL_83]] xor : f32
+    // CHECK:   [[VAL_98:%.*]], [[VAL_99:%.*]] = gpu.shuffle xor [[VAL_97]], [[VAL_8]], [[VAL_83]] : f32
     // CHECK:   cond_br [[VAL_99]], ^bb30, ^bb31
     // CHECK: ^bb30:
     // CHECK:   [[VAL_100:%.*]] = arith.cmpf ugt, [[VAL_97]], [[VAL_98]] : f32
     // CHECK:   [[VAL_101:%.*]] = select [[VAL_100]], [[VAL_97]], [[VAL_98]] : f32
     // CHECK:   br ^bb32([[VAL_101]] : f32)
     // CHECK: ^bb31:
     // CHECK:   br ^bb32([[VAL_97]] : f32)
     // CHECK: ^bb32([[VAL_102:%.*]]: f32):
-    // CHECK:   [[VAL_103:%.*]], [[VAL_104:%.*]] = gpu.shuffle [[VAL_102]], [[VAL_9]], [[VAL_83]] xor : f32
+    // CHECK:   [[VAL_103:%.*]], [[VAL_104:%.*]] = gpu.shuffle xor [[VAL_102]], [[VAL_9]], [[VAL_83]] : f32
     // CHECK:   cond_br [[VAL_104]], ^bb33, ^bb34
     // CHECK: ^bb33:
     // CHECK:   [[VAL_105:%.*]] = arith.cmpf ugt, [[VAL_102]], [[VAL_103]] : f32
     // CHECK:   [[VAL_106:%.*]] = select [[VAL_105]], [[VAL_102]], [[VAL_103]] : f32
     // CHECK:   br ^bb35([[VAL_106]] : f32)
     // CHECK: ^bb34:
     // CHECK:   br ^bb35([[VAL_102]] : f32)
     // CHECK: ^bb35([[VAL_107:%.*]]: f32):
-    // CHECK:   [[VAL_108:%.*]], [[VAL_109:%.*]] = gpu.shuffle [[VAL_107]], [[VAL_10]], [[VAL_83]] xor : f32
+    // CHECK:   [[VAL_108:%.*]], [[VAL_109:%.*]] = gpu.shuffle xor [[VAL_107]], [[VAL_10]], [[VAL_83]] : f32
     // CHECK:   cond_br [[VAL_109]], ^bb36, ^bb37
     // CHECK: ^bb36:
     // CHECK:   [[VAL_110:%.*]] = arith.cmpf ugt, [[VAL_107]], [[VAL_108]] : f32
     // CHECK:   [[VAL_111:%.*]] = select [[VAL_110]], [[VAL_107]], [[VAL_108]] : f32
     // CHECK:   br ^bb38([[VAL_111]] : f32)
     // CHECK: ^bb37:
     // CHECK:   br ^bb38([[VAL_107]] : f32)
     // CHECK: ^bb38([[VAL_112:%.*]]: f32):
     // CHECK:   br ^bb40([[VAL_112]] : f32)
     // CHECK: ^bb39:
-    // CHECK:   [[VAL_113:%.*]], [[VAL_114:%.*]] = gpu.shuffle [[VAL_86]], [[VAL_6]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_113:%.*]], [[VAL_114:%.*]] = gpu.shuffle xor [[VAL_86]], [[VAL_6]], [[VAL_5]] : f32
     // CHECK:   [[VAL_115:%.*]] = arith.cmpf ugt, [[VAL_86]], [[VAL_113]] : f32
     // CHECK:   [[VAL_116:%.*]] = select [[VAL_115]], [[VAL_86]], [[VAL_113]] : f32
-    // CHECK:   [[VAL_117:%.*]], [[VAL_118:%.*]] = gpu.shuffle [[VAL_116]], [[VAL_7]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_117:%.*]], [[VAL_118:%.*]] = gpu.shuffle xor [[VAL_116]], [[VAL_7]], [[VAL_5]] : f32
     // CHECK:   [[VAL_119:%.*]] = arith.cmpf ugt, [[VAL_116]], [[VAL_117]] : f32
     // CHECK:   [[VAL_120:%.*]] = select [[VAL_119]], [[VAL_116]], [[VAL_117]] : f32
-    // CHECK:   [[VAL_121:%.*]], [[VAL_122:%.*]] = gpu.shuffle [[VAL_120]], [[VAL_8]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_121:%.*]], [[VAL_122:%.*]] = gpu.shuffle xor [[VAL_120]], [[VAL_8]], [[VAL_5]] : f32
     // CHECK:   [[VAL_123:%.*]] = arith.cmpf ugt, [[VAL_120]], [[VAL_121]] : f32
     // CHECK:   [[VAL_124:%.*]] = select [[VAL_123]], [[VAL_120]], [[VAL_121]] : f32
-    // CHECK:   [[VAL_125:%.*]], [[VAL_126:%.*]] = gpu.shuffle [[VAL_124]], [[VAL_9]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_125:%.*]], [[VAL_126:%.*]] = gpu.shuffle xor [[VAL_124]], [[VAL_9]], [[VAL_5]] : f32
     // CHECK:   [[VAL_127:%.*]] = arith.cmpf ugt, [[VAL_124]], [[VAL_125]] : f32
     // CHECK:   [[VAL_128:%.*]] = select [[VAL_127]], [[VAL_124]], [[VAL_125]] : f32
-    // CHECK:   [[VAL_129:%.*]], [[VAL_130:%.*]] = gpu.shuffle [[VAL_128]], [[VAL_10]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_129:%.*]], [[VAL_130:%.*]] = gpu.shuffle xor [[VAL_128]], [[VAL_10]], [[VAL_5]] : f32
     // CHECK:   [[VAL_131:%.*]] = arith.cmpf ugt, [[VAL_128]], [[VAL_129]] : f32
     // CHECK:   [[VAL_132:%.*]] = select [[VAL_131]], [[VAL_128]], [[VAL_129]] : f32
     // CHECK:   br ^bb40([[VAL_132]] : f32)
     // CHECK: ^bb40([[VAL_133:%.*]]: f32):
     // CHECK:   store [[VAL_133]], [[VAL_1]]{{\[}}[[VAL_4]]] : memref<32xf32, 3>
     // CHECK:   br ^bb42
     // CHECK: ^bb41:
     // CHECK:   br ^bb42
     // CHECK: ^bb42:
     // CHECK:   gpu.barrier
-    %sum = "gpu.all_reduce"(%arg0) ({}) {op = "max"} : (f32) -> (f32)
+    %sum = gpu.all_reduce max %arg0 {} : (f32) -> (f32)
     gpu.return
   }
 
 }

mlir/test/Dialect/GPU/all-reduce.mlir

gpu.func @kernel(%arg0 : f32) kernel {
     // CHECK-DAG:   [[VAL_3:%.*]] = arith.constant 0 : i32
     // CHECK-DAG:   [[VAL_4:%.*]] = arith.constant 0 : index
     // CHECK-DAG:   [[VAL_5:%.*]] = arith.constant 32 : i32
     // CHECK-DAG:   [[VAL_6:%.*]] = arith.constant 1 : i32
     // CHECK-DAG:   [[VAL_7:%.*]] = arith.constant 2 : i32
     // CHECK-DAG:   [[VAL_8:%.*]] = arith.constant 4 : i32
     // CHECK-DAG:   [[VAL_9:%.*]] = arith.constant 8 : i32
     // CHECK-DAG:   [[VAL_10:%.*]] = arith.constant 16 : i32
-    // CHECK:   [[VAL_11:%.*]] = "gpu.block_dim"() {dimension = "x"} : () -> index
+    // CHECK:   [[VAL_11:%.*]] = gpu.block_dim x
     // CHECK:   [[VAL_12:%.*]] = arith.index_cast [[VAL_11]] : index to i32
-    // CHECK:   [[VAL_13:%.*]] = "gpu.block_dim"() {dimension = "y"} : () -> index
+    // CHECK:   [[VAL_13:%.*]] = gpu.block_dim y
     // CHECK:   [[VAL_14:%.*]] = arith.index_cast [[VAL_13]] : index to i32
-    // CHECK:   [[VAL_15:%.*]] = "gpu.block_dim"() {dimension = "z"} : () -> index
+    // CHECK:   [[VAL_15:%.*]] = gpu.block_dim z
     // CHECK:   [[VAL_16:%.*]] = arith.index_cast [[VAL_15]] : index to i32
-    // CHECK:   [[VAL_17:%.*]] = "gpu.thread_id"() {dimension = "x"} : () -> index
+    // CHECK:   [[VAL_17:%.*]] = gpu.thread_id x
     // CHECK:   [[VAL_18:%.*]] = arith.index_cast [[VAL_17]] : index to i32
-    // CHECK:   [[VAL_19:%.*]] = "gpu.thread_id"() {dimension = "y"} : () -> index
+    // CHECK:   [[VAL_19:%.*]] = gpu.thread_id y
     // CHECK:   [[VAL_20:%.*]] = arith.index_cast [[VAL_19]] : index to i32
-    // CHECK:   [[VAL_21:%.*]] = "gpu.thread_id"() {dimension = "z"} : () -> index
+    // CHECK:   [[VAL_21:%.*]] = gpu.thread_id z
     // CHECK:   [[VAL_22:%.*]] = arith.index_cast [[VAL_21]] : index to i32
     // CHECK:   [[VAL_23:%.*]] = arith.muli [[VAL_22]], [[VAL_14]] : i32
     // CHECK:   [[VAL_24:%.*]] = arith.addi [[VAL_23]], [[VAL_20]] : i32
     // CHECK:   [[VAL_25:%.*]] = arith.muli [[VAL_24]], [[VAL_12]] : i32
     // CHECK:   [[VAL_26:%.*]] = arith.muli [[VAL_12]], [[VAL_14]] : i32
     // CHECK:   [[VAL_27:%.*]] = arith.addi [[VAL_25]], [[VAL_18]] : i32
     // CHECK:   [[VAL_28:%.*]] = arith.muli [[VAL_26]], [[VAL_16]] : i32
     // CHECK:   [[VAL_29:%.*]] = arith.andi [[VAL_27]], [[VAL_2]] : i32
     // CHECK:   [[VAL_30:%.*]] = arith.cmpi eq, [[VAL_29]], [[VAL_3]] : i32
     // CHECK:   [[VAL_31:%.*]] = arith.subi [[VAL_27]], [[VAL_29]] : i32
     // CHECK:   [[VAL_32:%.*]] = arith.subi [[VAL_28]], [[VAL_31]] : i32
     // CHECK:   [[VAL_33:%.*]] = arith.cmpi slt, [[VAL_32]], [[VAL_5]] : i32
     // CHECK:   cond_br [[VAL_33]], ^bb1, ^bb17
     // CHECK: ^bb1:
-    // CHECK:   [[VAL_34:%.*]], [[VAL_35:%.*]] = gpu.shuffle [[VAL_0]], [[VAL_6]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_34:%.*]], [[VAL_35:%.*]] = gpu.shuffle xor [[VAL_0]], [[VAL_6]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_35]], ^bb2, ^bb3
     // CHECK: ^bb2:
     // CHECK:   [[VAL_36:%.*]] = arith.addf [[VAL_0]], [[VAL_34]] : f32
     // CHECK:   br ^bb4([[VAL_36]] : f32)
     // CHECK: ^bb3:
     // CHECK:   br ^bb4([[VAL_0]] : f32)
     // CHECK: ^bb4([[VAL_37:%.*]]: f32):
-    // CHECK:   [[VAL_38:%.*]], [[VAL_39:%.*]] = gpu.shuffle [[VAL_37]], [[VAL_7]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_38:%.*]], [[VAL_39:%.*]] = gpu.shuffle xor [[VAL_37]], [[VAL_7]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_39]], ^bb5, ^bb6
     // CHECK: ^bb5:
     // CHECK:   [[VAL_40:%.*]] = arith.addf [[VAL_37]], [[VAL_38]] : f32
     // CHECK:   br ^bb7([[VAL_40]] : f32)
     // CHECK: ^bb6:
     // CHECK:   br ^bb7([[VAL_37]] : f32)
     // CHECK: ^bb7([[VAL_41:%.*]]: f32):
-    // CHECK:   [[VAL_42:%.*]], [[VAL_43:%.*]] = gpu.shuffle [[VAL_41]], [[VAL_8]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_42:%.*]], [[VAL_43:%.*]] = gpu.shuffle xor [[VAL_41]], [[VAL_8]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_43]], ^bb8, ^bb9
     // CHECK: ^bb8:
     // CHECK:   [[VAL_44:%.*]] = arith.addf [[VAL_41]], [[VAL_42]] : f32
     // CHECK:   br ^bb10([[VAL_44]] : f32)
     // CHECK: ^bb9:
     // CHECK:   br ^bb10([[VAL_41]] : f32)
     // CHECK: ^bb10([[VAL_45:%.*]]: f32):
-    // CHECK:   [[VAL_46:%.*]], [[VAL_47:%.*]] = gpu.shuffle [[VAL_45]], [[VAL_9]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_46:%.*]], [[VAL_47:%.*]] = gpu.shuffle xor [[VAL_45]], [[VAL_9]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_47]], ^bb11, ^bb12
     // CHECK: ^bb11:
     // CHECK:   [[VAL_48:%.*]] = arith.addf [[VAL_45]], [[VAL_46]] : f32
     // CHECK:   br ^bb13([[VAL_48]] : f32)
     // CHECK: ^bb12:
     // CHECK:   br ^bb13([[VAL_45]] : f32)
     // CHECK: ^bb13([[VAL_49:%.*]]: f32):
-    // CHECK:   [[VAL_50:%.*]], [[VAL_51:%.*]] = gpu.shuffle [[VAL_49]], [[VAL_10]], [[VAL_32]] xor : f32
+    // CHECK:   [[VAL_50:%.*]], [[VAL_51:%.*]] = gpu.shuffle xor [[VAL_49]], [[VAL_10]], [[VAL_32]] : f32
     // CHECK:   cond_br [[VAL_51]], ^bb14, ^bb15
     // CHECK: ^bb14:
     // CHECK:   [[VAL_52:%.*]] = arith.addf [[VAL_49]], [[VAL_50]] : f32
     // CHECK:   br ^bb16([[VAL_52]] : f32)
     // CHECK: ^bb15:
     // CHECK:   br ^bb16([[VAL_49]] : f32)
     // CHECK: ^bb16([[VAL_53:%.*]]: f32):
     // CHECK:   br ^bb18([[VAL_53]] : f32)
     // CHECK: ^bb17:
-    // CHECK:   [[VAL_54:%.*]], [[VAL_55:%.*]] = gpu.shuffle [[VAL_0]], [[VAL_6]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_54:%.*]], [[VAL_55:%.*]] = gpu.shuffle xor [[VAL_0]], [[VAL_6]], [[VAL_5]] : f32
     // CHECK:   [[VAL_56:%.*]] = arith.addf [[VAL_0]], [[VAL_54]] : f32
-    // CHECK:   [[VAL_57:%.*]], [[VAL_58:%.*]] = gpu.shuffle [[VAL_56]], [[VAL_7]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_57:%.*]], [[VAL_58:%.*]] = gpu.shuffle xor [[VAL_56]], [[VAL_7]], [[VAL_5]] : f32
     // CHECK:   [[VAL_59:%.*]] = arith.addf [[VAL_56]], [[VAL_57]] : f32
-    // CHECK:   [[VAL_60:%.*]], [[VAL_61:%.*]] = gpu.shuffle [[VAL_59]], [[VAL_8]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_60:%.*]], [[VAL_61:%.*]] = gpu.shuffle xor [[VAL_59]], [[VAL_8]], [[VAL_5]] : f32
     // CHECK:   [[VAL_62:%.*]] = arith.addf [[VAL_59]], [[VAL_60]] : f32
-    // CHECK:   [[VAL_63:%.*]], [[VAL_64:%.*]] = gpu.shuffle [[VAL_62]], [[VAL_9]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_63:%.*]], [[VAL_64:%.*]] = gpu.shuffle xor [[VAL_62]], [[VAL_9]], [[VAL_5]] : f32
     // CHECK:   [[VAL_65:%.*]] = arith.addf [[VAL_62]], [[VAL_63]] : f32
-    // CHECK:   [[VAL_66:%.*]], [[VAL_67:%.*]] = gpu.shuffle [[VAL_65]], [[VAL_10]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_66:%.*]], [[VAL_67:%.*]] = gpu.shuffle xor [[VAL_65]], [[VAL_10]], [[VAL_5]] : f32
     // CHECK:   [[VAL_68:%.*]] = arith.addf [[VAL_65]], [[VAL_66]] : f32
     // CHECK:   br ^bb18([[VAL_68]] : f32)
     // CHECK: ^bb18([[VAL_69:%.*]]: f32):
     // CHECK:   cond_br [[VAL_30]], ^bb19, ^bb20
     // CHECK: ^bb19:
     // CHECK:   [[VAL_70:%.*]] = arith.divsi [[VAL_27]], [[VAL_5]] : i32
     // CHECK:   [[VAL_71:%.*]] = arith.index_cast [[VAL_70]] : i32 to index
     // CHECK:   store [[VAL_69]], [[VAL_1]]{{\[}}[[VAL_71]]] : memref<32xf32, 3>
     // CHECK:   br ^bb21
     // CHECK: ^bb20:
     // CHECK:   br ^bb21
     // CHECK: ^bb21:
     // CHECK:   gpu.barrier
     // CHECK:   [[VAL_72:%.*]] = arith.addi [[VAL_28]], [[VAL_2]] : i32
     // CHECK:   [[VAL_73:%.*]] = arith.divsi [[VAL_72]], [[VAL_5]] : i32
     // CHECK:   [[VAL_74:%.*]] = arith.cmpi slt, [[VAL_27]], [[VAL_73]] : i32
     // CHECK:   cond_br [[VAL_74]], ^bb22, ^bb41
     // CHECK: ^bb22:
     // CHECK:   [[VAL_75:%.*]] = arith.index_cast [[VAL_27]] : i32 to index
     // CHECK:   [[VAL_76:%.*]] = memref.load [[VAL_1]]{{\[}}[[VAL_75]]] : memref<32xf32, 3>
     // CHECK:   [[VAL_77:%.*]] = arith.cmpi slt, [[VAL_73]], [[VAL_5]] : i32
     // CHECK:   cond_br [[VAL_77]], ^bb23, ^bb39
     // CHECK: ^bb23:
-    // CHECK:   [[VAL_78:%.*]], [[VAL_79:%.*]] = gpu.shuffle [[VAL_76]], [[VAL_6]], [[VAL_73]] xor : f32
+    // CHECK:   [[VAL_78:%.*]], [[VAL_79:%.*]] = gpu.shuffle xor [[VAL_76]], [[VAL_6]], [[VAL_73]] : f32
     // CHECK:   cond_br [[VAL_79]], ^bb24, ^bb25
     // CHECK: ^bb24:
     // CHECK:   [[VAL_80:%.*]] = arith.addf [[VAL_76]], [[VAL_78]] : f32
     // CHECK:   br ^bb26([[VAL_80]] : f32)
     // CHECK: ^bb25:
     // CHECK:   br ^bb26([[VAL_76]] : f32)
     // CHECK: ^bb26([[VAL_81:%.*]]: f32):
-    // CHECK:   [[VAL_82:%.*]], [[VAL_83:%.*]] = gpu.shuffle [[VAL_81]], [[VAL_7]], [[VAL_73]] xor : f32
+    // CHECK:   [[VAL_82:%.*]], [[VAL_83:%.*]] = gpu.shuffle xor [[VAL_81]], [[VAL_7]], [[VAL_73]] : f32
     // CHECK:   cond_br [[VAL_83]], ^bb27, ^bb28
     // CHECK: ^bb27:
     // CHECK:   [[VAL_84:%.*]] = arith.addf [[VAL_81]], [[VAL_82]] : f32
     // CHECK:   br ^bb29([[VAL_84]] : f32)
     // CHECK: ^bb28:
     // CHECK:   br ^bb29([[VAL_81]] : f32)
     // CHECK: ^bb29([[VAL_85:%.*]]: f32):
-    // CHECK:   [[VAL_86:%.*]], [[VAL_87:%.*]] = gpu.shuffle [[VAL_85]], [[VAL_8]], [[VAL_73]] xor : f32
+    // CHECK:   [[VAL_86:%.*]], [[VAL_87:%.*]] = gpu.shuffle xor [[VAL_85]], [[VAL_8]], [[VAL_73]] : f32
     // CHECK:   cond_br [[VAL_87]], ^bb30, ^bb31
     // CHECK: ^bb30:
     // CHECK:   [[VAL_88:%.*]] = arith.addf [[VAL_85]], [[VAL_86]] : f32
     // CHECK:   br ^bb32([[VAL_88]] : f32)
     // CHECK: ^bb31:
     // CHECK:   br ^bb32([[VAL_85]] : f32)
     // CHECK: ^bb32([[VAL_89:%.*]]: f32):
-    // CHECK:   [[VAL_90:%.*]], [[VAL_91:%.*]] = gpu.shuffle [[VAL_89]], [[VAL_9]], [[VAL_73]] xor : f32
+    // CHECK:   [[VAL_90:%.*]], [[VAL_91:%.*]] = gpu.shuffle xor [[VAL_89]], [[VAL_9]], [[VAL_73]] : f32
     // CHECK:   cond_br [[VAL_91]], ^bb33, ^bb34
     // CHECK: ^bb33:
     // CHECK:   [[VAL_92:%.*]] = arith.addf [[VAL_89]], [[VAL_90]] : f32
     // CHECK:   br ^bb35([[VAL_92]] : f32)
     // CHECK: ^bb34:
     // CHECK:   br ^bb35([[VAL_89]] : f32)
     // CHECK: ^bb35([[VAL_93:%.*]]: f32):
-    // CHECK:   [[VAL_94:%.*]], [[VAL_95:%.*]] = gpu.shuffle [[VAL_93]], [[VAL_10]], [[VAL_73]] xor : f32
+    // CHECK:   [[VAL_94:%.*]], [[VAL_95:%.*]] = gpu.shuffle xor [[VAL_93]], [[VAL_10]], [[VAL_73]] : f32
     // CHECK:   cond_br [[VAL_95]], ^bb36, ^bb37
     // CHECK: ^bb36:
     // CHECK:   [[VAL_96:%.*]] = arith.addf [[VAL_93]], [[VAL_94]] : f32
     // CHECK:   br ^bb38([[VAL_96]] : f32)
     // CHECK: ^bb37:
     // CHECK:   br ^bb38([[VAL_93]] : f32)
     // CHECK: ^bb38([[VAL_97:%.*]]: f32):
     // CHECK:   br ^bb40([[VAL_97]] : f32)
     // CHECK: ^bb39:
-    // CHECK:   [[VAL_98:%.*]], [[VAL_99:%.*]] = gpu.shuffle [[VAL_76]], [[VAL_6]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_98:%.*]], [[VAL_99:%.*]] = gpu.shuffle xor [[VAL_76]], [[VAL_6]], [[VAL_5]] : f32
     // CHECK:   [[VAL_100:%.*]] = arith.addf [[VAL_76]], [[VAL_98]] : f32
-    // CHECK:   [[VAL_101:%.*]], [[VAL_102:%.*]] = gpu.shuffle [[VAL_100]], [[VAL_7]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_101:%.*]], [[VAL_102:%.*]] = gpu.shuffle xor [[VAL_100]], [[VAL_7]], [[VAL_5]] : f32
     // CHECK:   [[VAL_103:%.*]] = arith.addf [[VAL_100]], [[VAL_101]] : f32
-    // CHECK:   [[VAL_104:%.*]], [[VAL_105:%.*]] = gpu.shuffle [[VAL_103]], [[VAL_8]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_104:%.*]], [[VAL_105:%.*]] = gpu.shuffle xor [[VAL_103]], [[VAL_8]], [[VAL_5]] : f32
     // CHECK:   [[VAL_106:%.*]] = arith.addf [[VAL_103]], [[VAL_104]] : f32
-    // CHECK:   [[VAL_107:%.*]], [[VAL_108:%.*]] = gpu.shuffle [[VAL_106]], [[VAL_9]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_107:%.*]], [[VAL_108:%.*]] = gpu.shuffle xor [[VAL_106]], [[VAL_9]], [[VAL_5]] : f32
     // CHECK:   [[VAL_109:%.*]] = arith.addf [[VAL_106]], [[VAL_107]] : f32
-    // CHECK:   [[VAL_110:%.*]], [[VAL_111:%.*]] = gpu.shuffle [[VAL_109]], [[VAL_10]], [[VAL_5]] xor : f32
+    // CHECK:   [[VAL_110:%.*]], [[VAL_111:%.*]] = gpu.shuffle xor [[VAL_109]], [[VAL_10]], [[VAL_5]] : f32
     // CHECK:   [[VAL_112:%.*]] = arith.addf [[VAL_109]], [[VAL_110]] : f32
     // CHECK:   br ^bb40([[VAL_112]] : f32)
     // CHECK: ^bb40([[VAL_113:%.*]]: f32):
     // CHECK:   store [[VAL_113]], [[VAL_1]]{{\[}}[[VAL_4]]] : memref<32xf32, 3>
     // CHECK:   br ^bb42
     // CHECK: ^bb41:
     // CHECK:   br ^bb42
     // CHECK: ^bb42:
     // CHECK:   gpu.barrier
-    %sum = "gpu.all_reduce"(%arg0) ({}) {op = "add"} : (f32) -> (f32)
+    %sum = gpu.all_reduce add %arg0 {} : (f32) -> (f32)
     gpu.return
   }
 
 }

mlir/test/Dialect/GPU/invalid.mlir

func @launch_func_kernel_operand_attr(%sz : index) {
     // expected-error@+1 {{expected arguments without attributes}}
     gpu.launch_func @foo::@bar blocks in (%sz, %sz, %sz) threads in (%sz, %sz, %sz) args(%sz : index {foo})
     return
   }
 }
 
 // -----
 
-func @illegal_dimension() {
-  // expected-error@+1 {{dimension "o" is invalid}}
-  %tIdX = "gpu.thread_id"() {dimension = "o"} : () -> (index)
-
-  return
-}
-
-// -----
-
-func @illegal_dimension() {
-  // expected-error@+1 {{dimension "o" is invalid}}
-  %bDimX = "gpu.block_dim"() {dimension = "o"} : () -> (index)
-
-  return
-}
-
-// -----
-
-func @illegal_dimension() {
-  // expected-error@+1 {{dimension "o" is invalid}}
-  %bIdX = "gpu.block_id"() {dimension = "o"} : () -> (index)
-
-  return
-}
-
-// -----
-
-func @illegal_dimension() {
-  // expected-error@+1 {{dimension "o" is invalid}}
-  %gDimX = "gpu.grid_dim"() {dimension = "o"} : () -> (index)
-
-  return
-}
-
-// -----
-
 func @reduce_no_op_no_body(%arg0 : f32) {
   // expected-error@+1 {{expected either an op attribute or a non-empty body}}
   %res = "gpu.all_reduce"(%arg0) ({}) : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_op_and_body(%arg0 : f32) {
   // expected-error@+1 {{expected either an op attribute or a non-empty body}}
   %res = "gpu.all_reduce"(%arg0) ({
   ^bb(%lhs : f32, %rhs : f32):
     "gpu.yield"(%lhs) : (f32) -> ()
-  }) {op = "add"} : (f32) -> (f32)
+  }) {op = #gpu<"all_reduce_op add">} : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_invalid_op(%arg0 : f32) {
-  // expected-error@+1 {{attribute 'op' failed to satisfy constraint}}
-  %res = "gpu.all_reduce"(%arg0) ({}) {op = "foo"} : (f32) -> (f32)
+  // expected-error@+1 {{invalid op kind}}
+  %res = gpu.all_reduce foo %arg0 {} : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_invalid_op_type(%arg0 : f32) {
   // expected-error@+1 {{`and` accumulator is only compatible with Integer type}}
-  %res = "gpu.all_reduce"(%arg0) ({}) {op = "and"} : (f32) -> (f32)
+  %res = gpu.all_reduce and %arg0 {} : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_incorrect_region_arguments(%arg0 : f32) {
   // expected-error@+1 {{expected two region arguments}}
-  %res = "gpu.all_reduce"(%arg0) ({
+  %res = gpu.all_reduce %arg0 {
   ^bb(%lhs : f32):
     "gpu.yield"(%lhs) : (f32) -> ()
-  }) : (f32) -> (f32)
+  } : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_incorrect_region_arguments(%arg0 : f32) {
   // expected-error@+1 {{incorrect region argument type}}
-  %res = "gpu.all_reduce"(%arg0) ({
+  %res = gpu.all_reduce %arg0 {
   ^bb(%lhs : f32, %rhs : i32):
     "gpu.yield"(%lhs) : (f32) -> ()
-  }) : (f32) -> (f32)
+  } : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_incorrect_yield(%arg0 : f32) {
   // expected-error@+1 {{expected one gpu.yield operand}}
-  %res = "gpu.all_reduce"(%arg0) ({
+  %res = gpu.all_reduce %arg0 {
   ^bb(%lhs : f32, %rhs : f32):
     "gpu.yield"(%lhs, %rhs) : (f32, f32) -> ()
-  }) : (f32) -> (f32)
+  } : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_incorrect_yield(%arg0 : f32) {
   // expected-error@+1 {{incorrect gpu.yield type}}
-  %res = "gpu.all_reduce"(%arg0) ({
+  %res = gpu.all_reduce %arg0 {
   ^bb(%lhs : f32, %rhs : f32):
     %one = arith.constant 1 : i32
     "gpu.yield"(%one) : (i32) -> ()
-  }) : (f32) -> (f32)
+  } : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @reduce_incorrect_yield(%arg0 : f32) {
   // expected-error@+1 {{expected gpu.yield op in region}}
-  %res = "gpu.all_reduce"(%arg0) ({
+  %res = gpu.all_reduce %arg0 {
   ^bb(%lhs : f32, %rhs : f32):
     return
-  }) : (f32) -> (f32)
+  } : (f32) -> (f32)
   return
 }
 
 // -----
 
 func @shuffle_mismatching_type(%arg0 : f32, %arg1 : i32, %arg2 : i32) {
-  // expected-error@+1 {{requires the same type for value operand and result}}
-  %shfl, %pred = "gpu.shuffle"(%arg0, %arg1, %arg2) { mode = "xor" } : (f32, i32, i32) -> (i32, i1)
+  // expected-error@+1 {{inferred type(s) 'f32', 'i1' are incompatible with return type(s) of operation 'i32', 'i1'}}
+  %shfl, %pred = "gpu.shuffle"(%arg0, %arg1, %arg2) { mode = #gpu<"shuffle_mode xor"> } : (f32, i32, i32) -> (i32, i1)
+  return
 }
 
 // -----
 
 func @shuffle_unsupported_type(%arg0 : index, %arg1 : i32, %arg2 : i32) {
-  // expected-error@+1 {{requires value operand type to be f32 or i32}}
-  %shfl, %pred = gpu.shuffle %arg0, %arg1, %arg2 xor : index
+  // expected-error@+1 {{operand #0 must be i32 or f32}}
+  %shfl, %pred = gpu.shuffle xor %arg0, %arg1, %arg2 : index
+  return
 }
 
 // -----
 
 module {
   gpu.module @gpu_funcs {
     // expected-error @+1 {{custom op 'gpu.func' gpu.func requires named arguments}}
     gpu.func @kernel_1(f32, f32) {

mlir/test/Dialect/GPU/multiple-all-reduce.mlir

 // RUN: mlir-opt --gpu-kernel-outlining --convert-gpu-to-nvvm %s | FileCheck %s
 
 func @main() {
   %data = memref.alloc() : memref<2x6xf32>
   %sum = memref.alloc() : memref<2xf32>
   %mul = memref.alloc() : memref<2xf32>
   %c1 = arith.constant 1 : index
 
   // ADD + MUL
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c1, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %c1, %block_y = %c1, %block_z = %c1) {
     %val = memref.load %data[%bx, %tx] : memref<2x6xf32>
-    %reduced0 = "gpu.all_reduce"(%val) ({}) { op = "add" } : (f32) -> (f32)
+    %reduced0 = gpu.all_reduce add %val {} : (f32) -> (f32)
     memref.store %reduced0, %sum[%bx] : memref<2xf32>
-    %reduced1 = "gpu.all_reduce"(%val) ({}) { op = "mul" } : (f32) -> (f32)
+    %reduced1 = gpu.all_reduce mul %val {} : (f32) -> (f32)
     memref.store %reduced1, %mul[%bx] : memref<2xf32>
     gpu.terminator
   }
 
 // CHECK:      gpu.module @main_kernel {
 // CHECK-NEXT:   llvm.mlir.global internal @{{.*}}() {addr_space = 3 : i32} : !llvm.array<32 x f32>
 // CHECK-NEXT:   llvm.mlir.global internal @{{.*}}() {addr_space = 3 : i32} : !llvm.array<32 x f32>
 
   return
 }

mlir/test/Dialect/GPU/ops.mlir

gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %blk, %grid_y = %blk, %grid_z = %blk)
       // CHECK: gpu.terminator
       gpu.terminator
     }
     return
   }
 
   gpu.module @kernels {
     gpu.func @kernel_1(%arg0 : f32, %arg1 : memref<?xf32, 1>) kernel {
-      %tIdX = "gpu.thread_id"() {dimension = "x"} : () -> (index)
-      %tIdY = "gpu.thread_id"() {dimension = "y"} : () -> (index)
-      %tIdZ = "gpu.thread_id"() {dimension = "z"} : () -> (index)
-
-      %bDimX = "gpu.block_dim"() {dimension = "x"} : () -> (index)
-      %bDimY = "gpu.block_dim"() {dimension = "y"} : () -> (index)
-      %bDimZ = "gpu.block_dim"() {dimension = "z"} : () -> (index)
-
-      %bIdX = "gpu.block_id"() {dimension = "x"} : () -> (index)
-      %bIdY = "gpu.block_id"() {dimension = "y"} : () -> (index)
-      %bIdZ = "gpu.block_id"() {dimension = "z"} : () -> (index)
-
-      %gDimX = "gpu.grid_dim"() {dimension = "x"} : () -> (index)
-      %gDimY = "gpu.grid_dim"() {dimension = "y"} : () -> (index)
-      %gDimZ = "gpu.grid_dim"() {dimension = "z"} : () -> (index)
+      %tIdX = gpu.thread_id x
+      %tIdY = gpu.thread_id y
+      %tIdZ = gpu.thread_id z
+
+      %bDimX = gpu.block_dim x
+      %bDimY = gpu.block_dim y
+      %bDimZ = gpu.block_dim z
+
+      %bIdX = gpu.block_id x
+      %bIdY = gpu.block_id y
+      %bIdZ = gpu.block_id z
+
+      %gDimX = gpu.grid_dim x
+      %gDimY = gpu.grid_dim y
+      %gDimZ = gpu.grid_dim z
 
       %sgId = gpu.subgroup_id : index
       %numSg = gpu.num_subgroups : index
       %SgSi = gpu.subgroup_size : index
 
       %one = arith.constant 1.0 : f32
-      %sum = "gpu.all_reduce"(%one) ({}) {op = "add"} : (f32) -> (f32)
+      %sum = gpu.all_reduce add %one {} : (f32) -> (f32)
 
       %width = arith.constant 7 : i32
       %offset = arith.constant 3 : i32
-      // CHECK: gpu.shuffle %{{.*}}, %{{.*}}, %{{.*}} xor : f32
-      %shfl, %pred = gpu.shuffle %arg0, %offset, %width xor : f32
-      // CHECK: gpu.shuffle %{{.*}}, %{{.*}}, %{{.*}} up : f32
-      %shfl1, %pred1 = gpu.shuffle %arg0, %offset, %width up : f32
-      // CHECK: gpu.shuffle %{{.*}}, %{{.*}}, %{{.*}} down : f32
-      %shfl2, %pred2 = gpu.shuffle %arg0, %offset, %width down : f32
-      // CHECK: gpu.shuffle %{{.*}}, %{{.*}}, %{{.*}} idx : f32
-      %shfl3, %pred3 = gpu.shuffle %arg0, %offset, %width idx : f32
+      // CHECK: gpu.shuffle xor %{{.*}}, %{{.*}}, %{{.*}} : f32
+      %shfl, %pred = gpu.shuffle xor %arg0, %offset, %width : f32
+      // CHECK: gpu.shuffle up %{{.*}}, %{{.*}}, %{{.*}} : f32
+      %shfl1, %pred1 = gpu.shuffle up %arg0, %offset, %width : f32
+      // CHECK: gpu.shuffle down %{{.*}}, %{{.*}}, %{{.*}} : f32
+      %shfl2, %pred2 = gpu.shuffle down %arg0, %offset, %width : f32
+      // CHECK: gpu.shuffle idx %{{.*}}, %{{.*}}, %{{.*}} : f32
+      %shfl3, %pred3 = gpu.shuffle idx %arg0, %offset, %width : f32
 
       "gpu.barrier"() : () -> ()
 
       "some_op"(%bIdX, %tIdX) : (index, index) -> ()
       %42 = memref.load %arg1[%bIdX] : memref<?xf32, 1>
       gpu.return
     }
 
func @mmamatrix_valid_element_type(){
     // CHECK: %[[wg:.*]] = memref.alloca()
     %i = arith.constant 16 : index
     // CHECK: %[[i:.*]] = arith.constant 16 : index
      %cst = arith.constant 1.000000e+00 : f32
     // CHECK: %[[cst:.*]] = arith.constant 1.000000e+00 : f32
     %0 = gpu.subgroup_mma_load_matrix %wg[%i, %i] {leadDimension = 32 : index} : memref<32x32xf16, 3> -> !gpu.mma_matrix<16x16xf16, "AOp">
     // CHECK: gpu.subgroup_mma_load_matrix %[[wg]][%[[i]], %[[i]]] {leadDimension = 32 : index} : memref<32x32xf16, 3> -> !gpu.mma_matrix<16x16xf16, "AOp">
     %1 = gpu.subgroup_mma_constant_matrix %cst : !gpu.mma_matrix<16x16xf32, "COp">
-    // CHECK: gpu.subgroup_mma_elementwise %{{.*}}, %{{.*}} {operation = "ADDF"} : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
-    %2 = gpu.subgroup_mma_elementwise %1, %1 {operation = "ADDF"} : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
-    // CHECK: gpu.subgroup_mma_elementwise %{{.*}}, %{{.*}} {operation = "MAXF"} : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
-    %3 = gpu.subgroup_mma_elementwise %2, %1 {operation = "MAXF"} : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
+    // CHECK: gpu.subgroup_mma_elementwise addf %{{.*}}, %{{.*}} : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
+    %2 = gpu.subgroup_mma_elementwise addf %1, %1 : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
+    // CHECK: gpu.subgroup_mma_elementwise maxf %{{.*}}, %{{.*}} : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
+    %3 = gpu.subgroup_mma_elementwise maxf %2, %1 : (!gpu.mma_matrix<16x16xf32, "COp">, !gpu.mma_matrix<16x16xf32, "COp">) -> !gpu.mma_matrix<16x16xf32, "COp">
     return
   }
 }

mlir/test/Dialect/GPU/outlining.mlir

func @launch() {
   return
 }
 
 // CHECK-DL-LABEL: gpu.module @launch_kernel attributes {dlti.dl_spec = #dlti.dl_spec<#dlti.dl_entry<index, 32 : i32>>}
 
 // CHECK-LABEL: gpu.module @launch_kernel
 // CHECK-NEXT: gpu.func @launch_kernel
 // CHECK-SAME: (%[[KERNEL_ARG0:.*]]: f32, %[[KERNEL_ARG1:.*]]: memref<?xf32, 1>)
-// CHECK-NEXT: %[[BID:.*]] = "gpu.block_id"() {dimension = "x"} : () -> index
-// CHECK-NEXT: = "gpu.block_id"() {dimension = "y"} : () -> index
-// CHECK-NEXT: = "gpu.block_id"() {dimension = "z"} : () -> index
-// CHECK-NEXT: %[[TID:.*]] = "gpu.thread_id"() {dimension = "x"} : () -> index
-// CHECK-NEXT: = "gpu.thread_id"() {dimension = "y"} : () -> index
-// CHECK-NEXT: = "gpu.thread_id"() {dimension = "z"} : () -> index
-// CHECK-NEXT: = "gpu.grid_dim"() {dimension = "x"} : () -> index
-// CHECK-NEXT: = "gpu.grid_dim"() {dimension = "y"} : () -> index
-// CHECK-NEXT: = "gpu.grid_dim"() {dimension = "z"} : () -> index
-// CHECK-NEXT: %[[BDIM:.*]] = "gpu.block_dim"() {dimension = "x"} : () -> index
-// CHECK-NEXT: = "gpu.block_dim"() {dimension = "y"} : () -> index
-// CHECK-NEXT: = "gpu.block_dim"() {dimension = "z"} : () -> index
+// CHECK-NEXT: %[[BID:.*]] = gpu.block_id x
+// CHECK-NEXT: = gpu.block_id y
+// CHECK-NEXT: = gpu.block_id z
+// CHECK-NEXT: %[[TID:.*]] = gpu.thread_id x
+// CHECK-NEXT: = gpu.thread_id y
+// CHECK-NEXT: = gpu.thread_id z
+// CHECK-NEXT: = gpu.grid_dim x
+// CHECK-NEXT: = gpu.grid_dim y
+// CHECK-NEXT: = gpu.grid_dim z
+// CHECK-NEXT: %[[BDIM:.*]] = gpu.block_dim x
+// CHECK-NEXT: = gpu.block_dim y
+// CHECK-NEXT: = gpu.block_dim z
 // CHECK-NEXT: br ^[[BLOCK:.*]]
 // CHECK-NEXT: ^[[BLOCK]]:
 // CHECK-NEXT: "use"(%[[KERNEL_ARG0]]) : (f32) -> ()
 // CHECK-NEXT: "some_op"(%[[BID]], %[[BDIM]]) : (index, index) -> ()
 // CHECK-NEXT: = memref.load %[[KERNEL_ARG1]][%[[TID]]] : memref<?xf32, 1>
 
 // -----
 

mlir/test/Dialect/GPU/promotion.mlir

 // RUN: mlir-opt -allow-unregistered-dialect -test-gpu-memory-promotion -pass-pipeline='gpu.module(gpu.func(test-gpu-memory-promotion))' -split-input-file %s | FileCheck %s
 
 gpu.module @foo {
 
   // Verify that the attribution was indeed introduced
   // CHECK-LABEL: @memref3d
   // CHECK-SAME: (%[[arg:.*]]: memref<5x4xf32>
   // CHECK-SAME: workgroup(%[[promoted:.*]] : memref<5x4xf32, 3>)
   gpu.func @memref3d(%arg0: memref<5x4xf32> {gpu.test_promote_workgroup}) kernel {
     // Verify that loop bounds are emitted, the order does not matter.
     // CHECK-DAG: %[[c1:.*]] = arith.constant 1
     // CHECK-DAG: %[[c4:.*]] = arith.constant 4
     // CHECK-DAG: %[[c5:.*]] = arith.constant 5
-    // CHECK-DAG: %[[tx:.*]] = "gpu.thread_id"() {dimension = "x"}
-    // CHECK-DAG: %[[ty:.*]] = "gpu.thread_id"() {dimension = "y"}
-    // CHECK-DAG: %[[tz:.*]] = "gpu.thread_id"() {dimension = "z"}
-    // CHECK-DAG: %[[bdx:.*]] = "gpu.block_dim"() {dimension = "x"}
-    // CHECK-DAG: %[[bdy:.*]] = "gpu.block_dim"() {dimension = "y"}
-    // CHECK-DAG: %[[bdz:.*]] = "gpu.block_dim"() {dimension = "z"}
+    // CHECK-DAG: %[[tx:.*]] = gpu.thread_id x
+    // CHECK-DAG: %[[ty:.*]] = gpu.thread_id y
+    // CHECK-DAG: %[[tz:.*]] = gpu.thread_id z
+    // CHECK-DAG: %[[bdx:.*]] = gpu.block_dim x
+    // CHECK-DAG: %[[bdy:.*]] = gpu.block_dim y
+    // CHECK-DAG: %[[bdz:.*]] = gpu.block_dim z
 
     // Verify that loops for the copy are emitted. We only check the number of
     // loops here since their bounds are produced by mapLoopToProcessorIds,
     // tested separately.
     // CHECK: scf.for %[[i0:.*]] =
     // CHECK:   scf.for %[[i1:.*]] =
     // CHECK:     scf.for %[[i2:.*]] =
 
gpu.func @memref5d(%arg0: memref<8x7x6x5x4xf32> {gpu.test_promote_workgroup}) kernel {
     // Verify that loop bounds are emitted, the order does not matter.
     // CHECK-DAG: %[[c0:.*]] = arith.constant 0
     // CHECK-DAG: %[[c1:.*]] = arith.constant 1
     // CHECK-DAG: %[[c4:.*]] = arith.constant 4
     // CHECK-DAG: %[[c5:.*]] = arith.constant 5
     // CHECK-DAG: %[[c6:.*]] = arith.constant 6
     // CHECK-DAG: %[[c7:.*]] = arith.constant 7
     // CHECK-DAG: %[[c8:.*]] = arith.constant 8
-    // CHECK-DAG: %[[tx:.*]] = "gpu.thread_id"() {dimension = "x"}
-    // CHECK-DAG: %[[ty:.*]] = "gpu.thread_id"() {dimension = "y"}
-    // CHECK-DAG: %[[tz:.*]] = "gpu.thread_id"() {dimension = "z"}
-    // CHECK-DAG: %[[bdx:.*]] = "gpu.block_dim"() {dimension = "x"}
-    // CHECK-DAG: %[[bdy:.*]] = "gpu.block_dim"() {dimension = "y"}
-    // CHECK-DAG: %[[bdz:.*]] = "gpu.block_dim"() {dimension = "z"}
+    // CHECK-DAG: %[[tx:.*]] = gpu.thread_id x
+    // CHECK-DAG: %[[ty:.*]] = gpu.thread_id y
+    // CHECK-DAG: %[[tz:.*]] = gpu.thread_id z
+    // CHECK-DAG: %[[bdx:.*]] = gpu.block_dim x
+    // CHECK-DAG: %[[bdy:.*]] = gpu.block_dim y
+    // CHECK-DAG: %[[bdz:.*]] = gpu.block_dim z
 
     // Verify that loops for the copy are emitted.
     // CHECK: scf.for %[[i0:.*]] =
     // CHECK:   scf.for %[[i1:.*]] =
     // CHECK:     scf.for %[[i2:.*]] =
     // CHECK:       scf.for %[[i3:.*]] =
     // CHECK:         scf.for %[[i4:.*]] =
 

mlir/test/Dialect/LLVMIR/invalid.mlir

func @null_non_llvm_type() {
   // expected-error@+1 {{must be LLVM pointer type, but got 'i32'}}
   llvm.mlir.null : i32
 }
 
 // -----
 
 func @nvvm_invalid_shfl_pred_1(%arg0 : i32, %arg1 : i32, %arg2 : i32, %arg3 : i32) {
   // expected-error@+1 {{expected return type to be a two-element struct with i1 as the second element}}
-  %0 = nvvm.shfl.sync "bfly" %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> i32
+  %0 = nvvm.shfl.sync bfly %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> i32
 }
 
 // -----
 
 func @nvvm_invalid_shfl_pred_2(%arg0 : i32, %arg1 : i32, %arg2 : i32, %arg3 : i32) {
   // expected-error@+1 {{expected return type to be a two-element struct with i1 as the second element}}
-  %0 = nvvm.shfl.sync "bfly" %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32)>
+  %0 = nvvm.shfl.sync bfly %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32)>
 }
 
 // -----
 
 func @nvvm_invalid_shfl_pred_3(%arg0 : i32, %arg1 : i32, %arg2 : i32, %arg3 : i32) {
   // expected-error@+1 {{expected return type to be a two-element struct with i1 as the second element}}
-  %0 = nvvm.shfl.sync "bfly" %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i32)>
+  %0 = nvvm.shfl.sync bfly %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i32)>
 }
 
 // -----
 
 func @nvvm_invalid_mma_0(%a0 : f16, %a1 : vector<2xf16>,
                          %b0 : vector<2xf16>, %b1 : vector<2xf16>,
                          %c0 : f32, %c1 : f32, %c2 : f32, %c3 : f32,
                          %c4 : f32, %c5 : f32, %c6 : f32, %c7 : f32) {
module {
   }
 }
 
 // -----
 
 llvm.func @wmmaLoadOp_invalid_mem_space(%arg0: !llvm.ptr<i32, 5>, %arg1: i32) {
   // expected-error@+1 {{'nvvm.wmma.load' op expected source pointer in memory space 0, 1, 3}}
   %0 = nvvm.wmma.load %arg0, %arg1
-    {eltype = "f16", frag = "a", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+    {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<a>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (!llvm.ptr<i32, 5>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
   llvm.return
 }
 
 // -----
 
 llvm.func @wmmaLoadOp_invalid_AOp(%arg0: !llvm.ptr<i32, 3>, %arg1: i32) {
   // expected-error@+1 {{'nvvm.wmma.load' op expected destination type is a structure of 8 elements of type 'vector<2xf16>'}}
   %0 = nvvm.wmma.load %arg0, %arg1
-  {eltype = "f16", frag = "a", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+  {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<a>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
   : (!llvm.ptr<i32, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
   llvm.return
 }
 
 // -----
 
 llvm.func @wmmaLoadOp_invalid_BOp(%arg0: !llvm.ptr<i32, 3>, %arg1: i32) {
   // expected-error@+1 {{'nvvm.wmma.load' op expected destination type is a structure of 8 elements of type 'vector<2xf16>'}}
  %0 = nvvm.wmma.load %arg0, %arg1
- {eltype = "f16", frag = "b", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+ {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<b>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
  : (!llvm.ptr<i32, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 
   llvm.return
 }
 
 // -----
 
 llvm.func @wmmaLoadOp_invalid_COp(%arg0: !llvm.ptr<i32, 3>, %arg1: i32) {
   // expected-error@+1 {{'nvvm.wmma.load' op expected destination type is a structure of 4 elements of type 'vector<2xf16>'}}
  %0 = nvvm.wmma.load %arg0, %arg1
-   {eltype = "f16", frag = "c", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+   {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<c>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
    : (!llvm.ptr<i32, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>)>
 
   llvm.return
 }
 
 // -----
 
 llvm.func @wmmaStoreOp_invalid_mem_space(%arg0: !llvm.ptr<i32, 5>, %arg1: i32,
                             %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                             %arg4: vector<2 x f16>, %arg5: vector<2 xf16>) {
   // expected-error@+1 {{'nvvm.wmma.store' op expected operands to be a source pointer in memory space 0, 1, 3}}
   nvvm.wmma.store %arg0, %arg1, %arg2, %arg3, %arg4, %arg5
-    {eltype = "f16", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+    {eltype = #nvvm.mma_type<f16>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : !llvm.ptr<i32, 5>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>
   llvm.return
 }
 
 // -----
 
 llvm.func @gpu_wmma_mma_op_invalid_operands(%arg0: vector<2 x f16>, %arg1: vector<2 x f16>,
                         %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                         %arg4: vector<2 x f16>, %arg5: vector<2 x f16>,
                         %arg6: vector<2 x f16>, %arg7: vector<2 x f16>,
                         %arg8: vector<2 x f16>, %arg9: vector<2 x f16>,
                         %arg10: vector<2 x f16>, %arg11: vector<2 x f16>,
                         %arg12: vector<2 x f16>, %arg13: vector<2 x f16>,
                         %arg14: vector<2 x f16>, %arg15: vector<2 x f16>,
                         %arg16: vector<2 x f16>, %arg17: vector<2 x f16>,
                         %arg18: vector<2 x f16>) {
   // expected-error@+1 {{'nvvm.wmma.mma' op expected 20 arguments}}
   %0 = nvvm.wmma.mma %arg0, %arg1, %arg2, %arg3, %arg4, %arg5, %arg6, %arg7, %arg8, %arg9, %arg10, %arg11, %arg12, %arg13, %arg14, %arg15, %arg16, %arg17, %arg18
-    {eltypeA = "f16", eltypeB = "f16", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f16>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>,
        vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>)
       -> !llvm.struct<(vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>)>
   llvm.return
 }
 
 // -----
 
 llvm.func @gpu_wmma_mma_op_results(%arg0: vector<2 x f16>, %arg1: vector<2 x f16>,
                         %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                         %arg4: vector<2 x f16>, %arg5: vector<2 x f16>,
                         %arg6: vector<2 x f16>, %arg7: vector<2 x f16>,
                         %arg8: vector<2 x f16>, %arg9: vector<2 x f16>,
                         %arg10: vector<2 x f16>, %arg11: vector<2 x f16>,
                         %arg12: vector<2 x f16>, %arg13: vector<2 x f16>,
                         %arg14: vector<2 x f16>, %arg15: vector<2 x f16>,
                         %arg16: vector<2 x f16>, %arg17: vector<2 x f16>,
                         %arg18: vector<2 x f16>, %arg19: vector<2 x f16>) {
   // expected-error@+1 {{'nvvm.wmma.mma' op expected destination type is a structure of 4 elements of type 'vector<2xf16>'}}
   %0 = nvvm.wmma.mma %arg0, %arg1, %arg2, %arg3, %arg4, %arg5, %arg6, %arg7, %arg8, %arg9, %arg10, %arg11, %arg12, %arg13, %arg14, %arg15, %arg16, %arg17, %arg18, %arg19
-    {eltypeA = "f16", eltypeB = "f16", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f16>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>,
        vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>)
       -> !llvm.struct<(vector<2 x f16>, vector<2 x f16>, vector<2 x f16>)>  llvm.return
 }
 
 // -----
 
 llvm.func @gpu_wmma_mma_op_invalid_ab_operands(%arg0: vector<2 x f16>, %arg1: vector<2 x f16>,
                         %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                         %arg4: vector<2 x f16>, %arg5: vector<2 x f16>,
                         %arg6: vector<2 x f16>, %arg7: vector<2 x f16>,
                         %arg8: vector<2 x f16>, %arg9: vector<2 x f16>,
                         %arg10: vector<2 x f16>, %arg11: vector<2 x f16>,
                         %arg12: vector<2 x f16>, %arg13: vector<2 x f16>,
                         %arg14: vector<2 x f16>, %arg15: f32,
                         %arg16: f32, %arg17: f32, %arg18: f32, %arg19: f32,
                         %arg20: f32, %arg21: f32, %arg22: f32, %arg23: f32) {
   // expected-error@+1 {{'nvvm.wmma.mma' op expected argument 15 to be of type 'vector<2xf16>'}}
   %0 = nvvm.wmma.mma %arg0, %arg1, %arg2, %arg3, %arg4, %arg5, %arg6, %arg7, %arg8, %arg9, %arg10, %arg11, %arg12, %arg13, %arg14, %arg15, %arg16, %arg17, %arg18, %arg19, %arg20, %arg21, %arg22, %arg23
-    {eltypeA = "f16", eltypeB = "f32", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f32>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, f32, f32, f32, f32, f32, f32, f32, f32, f32) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
   llvm.return
 }
 
 // -----
 
 llvm.func @gpu_wmma_mma_op_invalid_c_operand(%arg0: vector<2 x f16>, %arg1: vector<2 x f16>,
                         %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                         %arg4: vector<2 x f16>, %arg5: vector<2 x f16>,
                         %arg6: vector<2 x f16>, %arg7: vector<2 x f16>,
                         %arg8: vector<2 x f16>, %arg9: vector<2 x f16>,
                         %arg10: vector<2 x f16>, %arg11: vector<2 x f16>,
                         %arg12: vector<2 x f16>, %arg13: vector<2 x f16>,
                         %arg14: vector<2 x f16>, %arg15: vector<2xf16>,
                         %arg16: f32, %arg17: f32, %arg18: f32, %arg19: f32,
                         %arg20: f32, %arg21: f32, %arg22: f32, %arg23: vector<2xf16>) {
   // expected-error@+1 {{'nvvm.wmma.mma' op expected argument 23 to be of type 'f32'}}
   %0 = nvvm.wmma.mma %arg0, %arg1, %arg2, %arg3, %arg4, %arg5, %arg6, %arg7, %arg8, %arg9, %arg10, %arg11, %arg12, %arg13, %arg14, %arg15, %arg16, %arg17, %arg18, %arg19, %arg20, %arg21, %arg22, %arg23
-    {eltypeA = "f16", eltypeB = "f32", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f32>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, f32, f32, f32, f32, f32, f32, f32, vector<2xf16>) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
   llvm.return
 }
 
 // -----
 
 llvm.func @gpu_wmma_mma_op_invalid_result(%arg0: vector<2 x f16>, %arg1: vector<2 x f16>,
                         %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                         %arg4: vector<2 x f16>, %arg5: vector<2 x f16>,
                         %arg6: vector<2 x f16>, %arg7: vector<2 x f16>,
                         %arg8: vector<2 x f16>, %arg9: vector<2 x f16>,
                         %arg10: vector<2 x f16>, %arg11: vector<2 x f16>,
                         %arg12: vector<2 x f16>, %arg13: vector<2 x f16>,
                         %arg14: vector<2 x f16>, %arg15: vector<2xf16>,
                         %arg16: f32, %arg17: f32, %arg18: f32, %arg19: f32,
                         %arg20: f32, %arg21: f32, %arg22: f32, %arg23: f32) {
   // expected-error@+1 {{'nvvm.wmma.mma' op expected destination type is a structure of 8 elements of type 'f32'}}
   %0 = nvvm.wmma.mma %arg0, %arg1, %arg2, %arg3, %arg4, %arg5, %arg6, %arg7, %arg8, %arg9, %arg10, %arg11, %arg12, %arg13, %arg14, %arg15, %arg16, %arg17, %arg18, %arg19, %arg20, %arg21, %arg22, %arg23
-    {eltypeA = "f16", eltypeB = "f32", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f32>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, f32, f32, f32, f32, f32, f32, f32, f32) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, vector<2xf16>)>
   llvm.return
 }
 
 // -----
 
 llvm.func @caller() {
   // expected-error @below {{expected function call to produce a value}}

mlir/test/Dialect/LLVMIR/nvvm.mlir

func @llvm.nvvm.barrier0() {
   // CHECK: nvvm.barrier0
   nvvm.barrier0
   llvm.return
 }
 
 func @nvvm_shfl(
     %arg0 : i32, %arg1 : i32, %arg2 : i32,
     %arg3 : i32, %arg4 : f32) -> i32 {
-  // CHECK: nvvm.shfl.sync "bfly" %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : i32 -> i32
-  %0 = nvvm.shfl.sync "bfly" %arg0, %arg3, %arg1, %arg2 : i32 -> i32
-  // CHECK: nvvm.shfl.sync "bfly" %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
-  %1 = nvvm.shfl.sync "bfly" %arg0, %arg4, %arg1, %arg2 : f32 -> f32
-  // CHECK: nvvm.shfl.sync "up" %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
-  %2 = nvvm.shfl.sync "up" %arg0, %arg4, %arg1, %arg2 : f32 -> f32
-  // CHECK: nvvm.shfl.sync "down" %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
-  %3 = nvvm.shfl.sync "down" %arg0, %arg4, %arg1, %arg2 : f32 -> f32
-  // CHECK: nvvm.shfl.sync "idx" %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
-  %4 = nvvm.shfl.sync "idx" %arg0, %arg4, %arg1, %arg2 : f32 -> f32
+  // CHECK: nvvm.shfl.sync bfly %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : i32 -> i32
+  %0 = nvvm.shfl.sync bfly %arg0, %arg3, %arg1, %arg2 : i32 -> i32
+  // CHECK: nvvm.shfl.sync bfly %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
+  %1 = nvvm.shfl.sync bfly %arg0, %arg4, %arg1, %arg2 : f32 -> f32
+  // CHECK: nvvm.shfl.sync up %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
+  %2 = nvvm.shfl.sync up %arg0, %arg4, %arg1, %arg2 : f32 -> f32
+  // CHECK: nvvm.shfl.sync down %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
+  %3 = nvvm.shfl.sync down %arg0, %arg4, %arg1, %arg2 : f32 -> f32
+  // CHECK: nvvm.shfl.sync idx %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : f32 -> f32
+  %4 = nvvm.shfl.sync idx %arg0, %arg4, %arg1, %arg2 : f32 -> f32
   llvm.return %0 : i32
 }
 
 func @nvvm_shfl_pred(
     %arg0 : i32, %arg1 : i32, %arg2 : i32,
     %arg3 : i32, %arg4 : f32) -> !llvm.struct<(i32, i1)> {
-  // CHECK: nvvm.shfl.sync "bfly" %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
-  %0 = nvvm.shfl.sync "bfly" %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
-  // CHECK: nvvm.shfl.sync "bfly" %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
-  %1 = nvvm.shfl.sync "bfly" %arg0, %arg4, %arg1, %arg2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+  // CHECK: nvvm.shfl.sync bfly %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
+  %0 = nvvm.shfl.sync bfly %arg0, %arg3, %arg1, %arg2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
+  // CHECK: nvvm.shfl.sync bfly %{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+  %1 = nvvm.shfl.sync bfly %arg0, %arg4, %arg1, %arg2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
   llvm.return %0 : !llvm.struct<(i32, i1)>
 }
 
 func @nvvm_vote(%arg0 : i32, %arg1 : i1) -> i32 {
   // CHECK: nvvm.vote.ballot.sync %{{.*}}, %{{.*}} : i32
   %0 = nvvm.vote.ballot.sync %arg0, %arg1 : i32
   llvm.return %0 : i32
 }
 
 func @nvvm_mma(%a0 : vector<2xf16>, %a1 : vector<2xf16>,
                %b0 : vector<2xf16>, %b1 : vector<2xf16>,
                %c0 : f32, %c1 : f32, %c2 : f32, %c3 : f32,
                %c4 : f32, %c5 : f32, %c6 : f32, %c7 : f32) {
   // CHECK: nvvm.mma.sync {{.*}} {alayout = "row", blayout = "col"} : (vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, f32, f32, f32, f32, f32, f32, f32, f32) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
-  %0 = nvvm.mma.sync %a0, %a1, %b0, %b1, %c0, %c1, %c2, %c3, %c4, %c5, %c6, %c7 {alayout="row", blayout="col"} : (vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, f32, f32, f32, f32, f32, f32, f32, f32) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
+  %0 = nvvm.mma.sync %a0, %a1, %b0, %b1, %c0, %c1, %c2, %c3, %c4, %c5, %c6, %c7 {alayout = "row", blayout = "col"} : (vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, f32, f32, f32, f32, f32, f32, f32, f32) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
   llvm.return %0 : !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
 }
 
 func @nvvm_wmma_load_tf32(%arg0: !llvm.ptr<i32>, %arg1 : i32) -> !llvm.struct<(i32, i32, i32, i32)> {
-  // CHECK: nvvm.wmma.load {{.*}} {eltype = "tf32", frag = "a", k = 8 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+  // CHECK: nvvm.wmma.load {{.*}} {eltype = #nvvm.mma_type<tf32>, frag = #nvvm.mma_frag<a>, k = 8 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
   %0 = nvvm.wmma.load %arg0, %arg1
-    {eltype = "tf32", frag = "a", k = 8 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+    {eltype = #nvvm.mma_type<tf32>, frag = #nvvm.mma_frag<a>, k = 8 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (!llvm.ptr<i32>) -> !llvm.struct<(i32, i32, i32, i32)>
   llvm.return %0 : !llvm.struct<(i32, i32, i32, i32)>
 }
 
 func @nvvm_wmma_mma(%0 : i32, %1 : i32, %2 : i32, %3 : i32, %4 : i32, %5 : i32,
                     %6 : i32, %7 : i32, %8 : f32, %9 : f32, %10 : f32,
                     %11 : f32, %12 : f32, %13 : f32, %14 : f32, %15 : f32)
                    -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)> {
-  // CHECK: nvvm.wmma.mma {{.*}} {eltypeA = "tf32", eltypeB = "f32", k = 8 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+  // CHECK: nvvm.wmma.mma {{.*}} {eltypeA = #nvvm.mma_type<tf32>, eltypeB = #nvvm.mma_type<f32>, k = 8 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
   %r = nvvm.wmma.mma %0, %1, %2, %3, %4, %5, %6, %7, %8, %9, %10, %11, %12, %13, %14, %15
-    {eltypeA = "tf32", eltypeB = "f32", k = 8 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<tf32>, eltypeB = #nvvm.mma_type<f32>, k = 8 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (i32, i32, i32, i32, i32, i32, i32, i32, f32, f32, f32, f32, f32, f32, f32, f32)
     -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
   llvm.return %r : !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
 }
 
 llvm.func @cp_async(%arg0: !llvm.ptr<i8, 3>, %arg1: !llvm.ptr<i8, 1>) {
 // CHECK:  nvvm.cp.async.shared.global %{{.*}}, %{{.*}}, 16
   nvvm.cp.async.shared.global %arg0, %arg1, 16

mlir/test/Dialect/Linalg/distribute-tiled-loop.mlir

 }
 
 // CHECK-DAG: #[[$MAP0:.+]] = affine_map<()[s0] -> (s0 * 24)>
 // CHECK-DAG: #[[$MAP1:.+]] = affine_map<()[s0] -> (s0 * 16)>
 
 // CHECK-LABEL: func @distribute_for_gpu
 // CHECK:  %[[C64:.*]] = arith.constant 64 : index
 
-// CHECK-DAG:  %[[GPU_BLOCK_X:.*]] = "gpu.block_id"() {dimension = "x"}
-// CHECK-DAG:  %[[GPU_GRID_DIM_X:.*]] = "gpu.grid_dim"() {dimension = "x"}
+// CHECK-DAG:  %[[GPU_BLOCK_X:.*]] = gpu.block_id x
+// CHECK-DAG:  %[[GPU_GRID_DIM_X:.*]] = gpu.grid_dim x
 // CHECK-DAG:  %[[LB_I:.*]] = affine.apply #[[$MAP0]](){{\[}}%[[GPU_BLOCK_X]]]
 // CHECK-DAG:  %[[STEP_I:.*]] = affine.apply #[[$MAP0]](){{\[}}%[[GPU_GRID_DIM_X]]]
 
-// CHECK-DAG:  %[[GPU_BLOCK_Y:.*]] = "gpu.block_id"() {dimension = "y"}
-// CHECK-DAG:  %[[GPU_GRID_DIM_Y:.*]] = "gpu.grid_dim"() {dimension = "y"}
+// CHECK-DAG:  %[[GPU_BLOCK_Y:.*]] = gpu.block_id y
+// CHECK-DAG:  %[[GPU_GRID_DIM_Y:.*]] = gpu.grid_dim y
 // CHECK-DAG:  %[[LB_J:.*]] = affine.apply #[[$MAP1]](){{\[}}%[[GPU_BLOCK_Y]]]
 // CHECK-DAG:  %[[STEP_J:.*]] = affine.apply #[[$MAP1]](){{\[}}%[[GPU_GRID_DIM_Y]]]
 
 // CHECK:  linalg.tiled_loop (%[[I:.*]], %[[J:.*]]) = (%[[LB_I]], %[[LB_J]])
 // CHECK-SAME: to (%[[C64]], %[[C64]]) step (%[[STEP_I]], %[[STEP_J]])

mlir/test/Dialect/Linalg/tile-and-distribute.mlir

 // RUN: mlir-opt %s -test-linalg-transform-patterns=test-tile-and-distribute-options -split-input-file | FileCheck %s
 
 func @gemm1(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
 {
   linalg.matmul {__internal_linalg_transform__ = "distribute1"}
     ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
    outs(%c: memref<?x?xf32>)
   return
 }
 //  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
 //      CHECK: func @gemm1(
 // CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = "gpu.block_id"() {dimension = "y"}
-//  CHECK-DAG: %[[BIDX:.*]] = "gpu.block_id"() {dimension = "x"}
+//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
+//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
 //      CHECK: scf.for %[[ARG3:.*]] =
 //      CHECK:   %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
 //      CHECK:   %[[SV1:.*]] = memref.subview %[[ARG0]][%[[OFFSETY]], %[[ARG3]]]
 //      CHECK:   %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK:   %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG3]], %[[OFFSETX]]]
 //      CHECK:   %[[OFFSETY_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
 //      CHECK:   %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK:   %[[SV3:.*]] = memref.subview %[[ARG2]][%[[OFFSETY_2]], %[[OFFSETX]]]
 //      CHECK:   linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
 
 // -----
 
 func @gemm2(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
 {
   linalg.matmul  {__internal_linalg_transform__ = "distribute2"}
     ins(%a, %b: memref<?x?xf32>, memref<?x?xf32>)
    outs(%c:memref<?x?xf32>)
   return
 }
 //  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
 //      CHECK: func @gemm2(
 // CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = "gpu.block_id"() {dimension = "y"}
-//  CHECK-DAG: %[[BIDX:.*]] = "gpu.block_id"() {dimension = "x"}
+//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
+//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
 //      CHECK: %[[ITERY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
 //      CHECK: %[[ITERX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK: %[[INBOUNDSY:.*]] = arith.cmpi slt, %[[ITERY]], %{{.*}}
 //      CHECK: %[[INBOUNDSX:.*]] = arith.cmpi slt, %[[ITERX]], %{{.*}}
 //      CHECK: %[[INBOUNDS:.*]] = arith.andi %[[INBOUNDSY]], %[[INBOUNDSX]]
 //      CHECK: scf.if %[[INBOUNDS]]
 //      CHECK:   scf.for %[[ARG3:.*]] =
 //      CHECK:     %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
func @gemm3(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
    outs(%c: memref<?x?xf32>)
   return
 }
 //  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
 //      CHECK: func @gemm3(
 // CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = "gpu.block_id"() {dimension = "y"}
-//  CHECK-DAG: %[[NBLOCKSY:.*]] = "gpu.grid_dim"() {dimension = "y"}
-//  CHECK-DAG: %[[BIDX:.*]] = "gpu.block_id"() {dimension = "x"}
-//  CHECK-DAG: %[[NBLOCKSX:.*]] = "gpu.grid_dim"() {dimension = "x"}
+//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
+//  CHECK-DAG: %[[NBLOCKSY:.*]] = gpu.grid_dim y
+//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
+//  CHECK-DAG: %[[NBLOCKSX:.*]] = gpu.grid_dim x
 //      CHECK: %[[LBY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
 //      CHECK: %[[STEPY:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSY]]]
 //      CHECK: %[[LBX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK: %[[STEPX:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSX]]]
 //      CHECK: scf.parallel (%[[ARG3:.*]], %[[ARG4:.*]]) = (%[[LBY]], %[[LBX]]) to (%{{.*}}, %{{.*}}) step (%[[STEPY]], %[[STEPX]])
 //      CHECK:   scf.for %[[ARG5:.*]] =
 //      CHECK:     %[[SV1:.*]] = memref.subview %[[ARG0]][%[[ARG3]], %[[ARG5]]]
 //      CHECK:     %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG5]], %[[ARG4]]]
func @gemm4(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
    outs(%c: memref<?x?xf32>)
   return
 }
 //  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
 //      CHECK: func @gemm4(
 // CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = "gpu.block_id"() {dimension = "y"}
-//  CHECK-DAG: %[[BIDX:.*]] = "gpu.block_id"() {dimension = "x"}
+//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
+//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
 //      CHECK: %[[LBX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK: %[[INBOUNDS:.*]] = arith.cmpi slt, %[[LBX]], %{{.*}}
 //      CHECK: scf.if %[[INBOUNDS]]
 //      CHECK:   scf.for %[[ARG3:.*]] =
 //      CHECK:     %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
 //      CHECK:     %[[SV1:.*]] = memref.subview %[[ARG0]][%[[OFFSETY]], %[[ARG3]]]
 //      CHECK:     %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK:     %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG3]], %[[OFFSETX]]]
func @gemm5(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
    outs(%c: memref<?x?xf32>)
   return
 }
 //  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
 //      CHECK: func @gemm5(
 // CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = "gpu.block_id"() {dimension = "y"}
-//  CHECK-DAG: %[[BIDX:.*]] = "gpu.block_id"() {dimension = "x"}
-//  CHECK-DAG: %[[NBLOCKSX:.*]] = "gpu.grid_dim"() {dimension = "x"}
+//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
+//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
+//  CHECK-DAG: %[[NBLOCKSX:.*]] = gpu.grid_dim x
 //      CHECK: %[[LBY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
 //      CHECK: %[[LBX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK: %[[STEPX:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSX]]]
 //      CHECK: %[[INBOUNDS:.*]] = arith.cmpi slt, %[[LBY]], %{{.*}}
 //      CHECK: scf.if %[[INBOUNDS]]
 //      CHECK:   scf.parallel (%[[ARG3:.*]]) = (%[[LBX]]) to (%{{.*}}) step (%[[STEPX]])
 //      CHECK:     scf.for %[[ARG4:.*]] =
 //      CHECK:      %[[OFFSETY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
func @gemm6(%a : memref<?x?xf32>, %b : memref<?x?xf32>, %c : memref<?x?xf32>)
    outs(%c: memref<?x?xf32>)
   return
 }
 //  CHECK-DAG: #[[MAP0:.*]] = affine_map<()[s0] -> (s0 * 8)>
 //      CHECK: func @gemm6(
 // CHECK-SAME: %[[ARG0:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG1:[a-zA-Z0-9_]*]]: memref<?x?xf32>
 // CHECK-SAME: %[[ARG2:[a-zA-Z0-9_]*]]: memref<?x?xf32>
-//  CHECK-DAG: %[[BIDY:.*]] = "gpu.block_id"() {dimension = "y"}
-//  CHECK-DAG: %[[NBLOCKSY:.*]] = "gpu.grid_dim"() {dimension = "y"}
-//  CHECK-DAG: %[[BIDX:.*]] = "gpu.block_id"() {dimension = "x"}
+//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
+//  CHECK-DAG: %[[NBLOCKSY:.*]] = gpu.grid_dim y
+//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
 //      CHECK: %[[LBY:.*]] = affine.apply #[[MAP0]]()[%[[BIDY]]]
 //      CHECK: %[[STEPY:.*]] = affine.apply #[[MAP0]]()[%[[NBLOCKSY]]]
 //      CHECK: scf.parallel (%[[ARG3:.*]]) = (%[[LBY]]) to (%{{.*}}) step (%[[STEPY]])
 //      CHECK:   scf.for %[[ARG4:.*]] =
 //      CHECK:     %[[SV1:.*]] = memref.subview %[[ARG0]][%[[ARG3]], %[[ARG4]]]
 //      CHECK:     %[[OFFSETX:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK:     %[[SV2:.*]] = memref.subview %[[ARG1]][%[[ARG4]], %[[OFFSETX]]]
 //      CHECK:     %[[OFFSETX_2:.*]] = affine.apply #[[MAP0]]()[%[[BIDX]]]
 //      CHECK:     %[[SV3:.*]] = memref.subview %[[ARG2]][%[[ARG3]], %[[OFFSETX_2]]]
 //      CHECK:     linalg.matmul ins(%[[SV1]], %[[SV2]]{{.*}} outs(%[[SV3]]
 
 // -----
 
 //      CHECK: #[[MULMAP:.+]] = affine_map<()[s0, s1] -> (s0 * s1)>
 //      CHECK: #[[ADDMAP:.+]] = affine_map<()[s0, s1] -> (s0 + s1)>
 //      CHECK: func @matmul_tensors(
 // CHECK-SAME:    %[[TA:[0-9a-z]+]]: tensor<?x?xf32>
 // CHECK-SAME:    %[[TB:[0-9a-z]+]]: tensor<?x?xf32>
 // CHECK-SAME:    %[[TC:[0-9a-z]+]]: tensor<?x?xf32>) -> tensor<?x?xf32> {
 func @matmul_tensors(
   %arg0: tensor<?x?xf32>, %arg1: tensor<?x?xf32>, %arg2: tensor<?x?xf32>)
     -> tensor<?x?xf32> {
 //  CHECK-DAG: %[[C8:.*]] = arith.constant 8 : index
 //  CHECK-DAG: %[[C0:.*]] = arith.constant 0 : index
-//  CHECK-DAG: %[[BIDY:.*]] = "gpu.block_id"() {dimension = "y"}
-//  CHECK-DAG: %[[NBLOCKSY:.*]] = "gpu.grid_dim"() {dimension = "y"}
-//  CHECK-DAG: %[[BIDX:.*]] = "gpu.block_id"() {dimension = "x"}
-//  CHECK-DAG: %[[NBLOCKSX:.*]] = "gpu.grid_dim"() {dimension = "x"}
+//  CHECK-DAG: %[[BIDY:.*]] = gpu.block_id y
+//  CHECK-DAG: %[[NBLOCKSY:.*]] = gpu.grid_dim y
+//  CHECK-DAG: %[[BIDX:.*]] = gpu.block_id x
+//  CHECK-DAG: %[[NBLOCKSX:.*]] = gpu.grid_dim x
 //      CHECK: %[[MUL:.+]] = affine.apply #[[MULMAP]]()[%[[BIDY]], %[[C8]]]
 //      CHECK: %[[LBY:.+]] = affine.apply #[[ADDMAP]]()[%[[MUL]], %[[C0]]]
 //      CHECK: %[[STEPY:.+]] = affine.apply #[[MULMAP]]()[%[[NBLOCKSY]], %[[C8]]]
 //      CHECK: %[[TD0:.*]] = scf.for {{.*}} to {{.*}} step {{.*}} iter_args(%[[TC0:.*]] = %[[TC]]) -> (tensor<?x?xf32>) {
 //      CHECK: %[[MUL:.+]] = affine.apply #[[MULMAP]]()[%[[BIDX]], %[[C8]]]
 //      CHECK: %[[LBX:.+]] = affine.apply #[[ADDMAP]]()[%[[MUL]], %[[C0]]]
 //      CHECK: %[[STEPX:.+]] = affine.apply #[[MULMAP]]()[%[[NBLOCKSX]], %[[C8]]]
 //      CHECK:   %[[TD1:.*]] = scf.for {{.*}} to {{.*}} step {{.*}} iter_args(%[[TC1:.*]] = %[[TC0]]) -> (tensor<?x?xf32>) {

mlir/test/Dialect/OpenMP/invalid.mlir

func @omp_ordered2(%arg1 : i32, %arg2 : i32, %arg3 : i32) -> () {
   }
   return
 }
 
 // -----
 
 func @omp_ordered3(%vec0 : i64) -> () {
   // expected-error @below {{ordered depend directive must be closely nested inside a worksharing-loop with ordered clause with parameter present}}
-  omp.ordered depend_type("dependsink") depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
+  omp.ordered depend_type(dependsink) depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
   return
 }
 
 // -----
 
 func @omp_ordered4(%arg1 : i32, %arg2 : i32, %arg3 : i32, %vec0 : i64) -> () {
   omp.wsloop (%0) : i32 = (%arg1) to (%arg2) step (%arg3) ordered(0) {
     // expected-error @below {{ordered depend directive must be closely nested inside a worksharing-loop with ordered clause with parameter present}}
-    omp.ordered depend_type("dependsink") depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
+    omp.ordered depend_type(dependsink) depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
 
     omp.yield
   }
   return
 }
 // -----
 
 func @omp_ordered5(%arg1 : i32, %arg2 : i32, %arg3 : i32, %vec0 : i64, %vec1 : i64) -> () {
   omp.wsloop (%0) : i32 = (%arg1) to (%arg2) step (%arg3) ordered(1) {
     // expected-error @below {{number of variables in depend clause does not match number of iteration variables in the doacross loop}}
-    omp.ordered depend_type("dependsource") depend_vec(%vec0, %vec1 : i64, i64) {num_loops_val = 2 : i64}
+    omp.ordered depend_type(dependsource) depend_vec(%vec0, %vec1 : i64, i64) {num_loops_val = 2 : i64}
 
     omp.yield
   }
   return
 }
 
 // -----
 

mlir/test/Dialect/OpenMP/ops.mlir

 func @omp_parallel(%data_var : memref<i32>, %if_cond : i1, %num_threads : si32) -> () {
   // CHECK: omp.parallel if(%{{.*}}) num_threads(%{{.*}} : si32) private(%{{.*}} : memref<i32>) firstprivate(%{{.*}} : memref<i32>) shared(%{{.*}} : memref<i32>) copyin(%{{.*}} : memref<i32>) allocate(%{{.*}} : memref<i32> -> %{{.*}} : memref<i32>)
   "omp.parallel" (%if_cond, %num_threads, %data_var, %data_var, %data_var, %data_var, %data_var, %data_var) ({
 
   // test without if condition
   // CHECK: omp.parallel num_threads(%{{.*}} : si32) private(%{{.*}} : memref<i32>) firstprivate(%{{.*}} : memref<i32>) shared(%{{.*}} : memref<i32>) copyin(%{{.*}} : memref<i32>) allocate(%{{.*}} : memref<i32> -> %{{.*}} : memref<i32>)
     "omp.parallel"(%num_threads, %data_var, %data_var, %data_var, %data_var, %data_var, %data_var) ({
       omp.terminator
-    }) {operand_segment_sizes = dense<[0,1,1,1,1,1,1,1]>: vector<8xi32>, default_val = "defshared"} : (si32, memref<i32>, memref<i32>, memref<i32>, memref<i32>, memref<i32>, memref<i32>) -> ()
+    }) {operand_segment_sizes = dense<[0,1,1,1,1,1,1,1]>: vector<8xi32>, default_val = #omp<"clause_default defshared">} : (si32, memref<i32>, memref<i32>, memref<i32>, memref<i32>, memref<i32>, memref<i32>) -> ()
 
   // CHECK: omp.barrier
     omp.barrier
 
   // test without num_threads
   // CHECK: omp.parallel if(%{{.*}}) private(%{{.*}} : memref<i32>) firstprivate(%{{.*}} : memref<i32>) shared(%{{.*}} : memref<i32>) copyin(%{{.*}} : memref<i32>) allocate(%{{.*}} : memref<i32> -> %{{.*}} : memref<i32>)
     "omp.parallel"(%if_cond, %data_var, %data_var, %data_var, %data_var, %data_var, %data_var) ({
       omp.terminator
omp.wsloop (%0) : i32 = (%arg1) to (%arg2) step (%arg3) ordered(0) {
     omp.ordered_region {
       omp.terminator
     }
     omp.yield
   }
 
   omp.wsloop (%0) : i32 = (%arg1) to (%arg2) step (%arg3) ordered(1) {
     // Only one DEPEND(SINK: vec) clause
-    // CHECK: omp.ordered depend_type("dependsink") depend_vec(%{{.*}} : i64) {num_loops_val = 1 : i64}
-    omp.ordered depend_type("dependsink") depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
+    // CHECK: omp.ordered depend_type(dependsink) depend_vec(%{{.*}} : i64) {num_loops_val = 1 : i64}
+    omp.ordered depend_type(dependsink) depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
 
-    // CHECK: omp.ordered depend_type("dependsource") depend_vec(%{{.*}} : i64) {num_loops_val = 1 : i64}
-    omp.ordered depend_type("dependsource") depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
+    // CHECK: omp.ordered depend_type(dependsource) depend_vec(%{{.*}} : i64) {num_loops_val = 1 : i64}
+    omp.ordered depend_type(dependsource) depend_vec(%vec0 : i64) {num_loops_val = 1 : i64}
 
     omp.yield
   }
 
   omp.wsloop (%0) : i32 = (%arg1) to (%arg2) step (%arg3) ordered(2) {
     // Multiple DEPEND(SINK: vec) clauses
-    // CHECK: omp.ordered depend_type("dependsink") depend_vec(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : i64, i64, i64, i64) {num_loops_val = 2 : i64}
-    omp.ordered depend_type("dependsink") depend_vec(%vec0, %vec1, %vec2, %vec3 : i64, i64, i64, i64) {num_loops_val = 2 : i64}
+    // CHECK: omp.ordered depend_type(dependsink) depend_vec(%{{.*}}, %{{.*}}, %{{.*}}, %{{.*}} : i64, i64, i64, i64) {num_loops_val = 2 : i64}
+    omp.ordered depend_type(dependsink) depend_vec(%vec0, %vec1, %vec2, %vec3 : i64, i64, i64, i64) {num_loops_val = 2 : i64}
 
-    // CHECK: omp.ordered depend_type("dependsource") depend_vec(%{{.*}}, %{{.*}} : i64, i64) {num_loops_val = 2 : i64}
-    omp.ordered depend_type("dependsource") depend_vec(%vec0, %vec1 : i64, i64) {num_loops_val = 2 : i64}
+    // CHECK: omp.ordered depend_type(dependsource) depend_vec(%{{.*}}, %{{.*}} : i64, i64) {num_loops_val = 2 : i64}
+    omp.ordered depend_type(dependsource) depend_vec(%vec0, %vec1 : i64, i64) {num_loops_val = 2 : i64}
 
     omp.yield
   }
 
   return
 }
 
 // CHECK-LABEL: omp_atomic_read

mlir/test/Integration/GPU/CUDA/all-reduce-and.mlir

func @main() {
   memref.store %cst7, %data[%c1, %c3] : memref<2x6xi32>
   memref.store %cst10, %data[%c1, %c4] : memref<2x6xi32>
   memref.store %cst11, %data[%c1, %c5] : memref<2x6xi32>
 
   // AND
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c2, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %c6, %block_y = %c1, %block_z = %c1) {
     %val = memref.load %data[%bx, %tx] : memref<2x6xi32>
-    %reduced = "gpu.all_reduce"(%val) ({}) { op = "and" } : (i32) -> (i32)
+    %reduced = gpu.all_reduce and %val {} : (i32) -> (i32)
     memref.store %reduced, %sum[%bx] : memref<2xi32>
     gpu.terminator
   }
 
   call @print_memref_i32(%cast_sum) : (memref<*xi32>) -> ()
   // CHECK: [0, 2]
 
   return
 }
 
 func private @print_memref_i32(memref<*xi32>)
 

mlir/test/Integration/GPU/CUDA/all-reduce-max.mlir

func @main() {
   memref.store %cst7, %data[%c1, %c3] : memref<2x6xi32>
   memref.store %cst10, %data[%c1, %c4] : memref<2x6xi32>
   memref.store %cst11, %data[%c1, %c5] : memref<2x6xi32>
 
   // MAX
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c2, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %c6, %block_y = %c1, %block_z = %c1) {
     %val = memref.load %data[%bx, %tx] : memref<2x6xi32>
-    %reduced = "gpu.all_reduce"(%val) ({}) { op = "max" } : (i32) -> (i32)
+    %reduced = gpu.all_reduce max %val {} : (i32) -> (i32)
     memref.store %reduced, %sum[%bx] : memref<2xi32>
     gpu.terminator
   }
 
   call @print_memref_i32(%cast_sum) : (memref<*xi32>) -> ()
   // CHECK: [16, 11]
 
   return
 }
 
 func private @print_memref_i32(memref<*xi32>)
 

mlir/test/Integration/GPU/CUDA/all-reduce-min.mlir

func @main() {
   memref.store %cst7, %data[%c1, %c3] : memref<2x6xi32>
   memref.store %cst10, %data[%c1, %c4] : memref<2x6xi32>
   memref.store %cst11, %data[%c1, %c5] : memref<2x6xi32>
 
   // MIN
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c2, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %c6, %block_y = %c1, %block_z = %c1) {
     %val = memref.load %data[%bx, %tx] : memref<2x6xi32>
-    %reduced = "gpu.all_reduce"(%val) ({}) { op = "min" } : (i32) -> (i32)
+    %reduced = gpu.all_reduce min %val {} : (i32) -> (i32)
     memref.store %reduced, %sum[%bx] : memref<2xi32>
     gpu.terminator
   }
 
   call @print_memref_i32(%cast_sum) : (memref<*xi32>) -> ()
   // CHECK: [0, 2]
 
   return
 }
 
 func private @print_memref_i32(memref<*xi32>)
 

mlir/test/Integration/GPU/CUDA/all-reduce-op.mlir

func @main() {
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c1, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %sx, %block_y = %sy, %block_z = %sz) {
     %t0 = arith.muli %tz, %block_y : index
     %t1 = arith.addi %ty, %t0 : index
     %t2 = arith.muli %t1, %block_x : index
     %idx = arith.addi %tx, %t2 : index
     %t3 = arith.index_cast %idx : index to i32
     %val = arith.sitofp %t3 : i32 to f32
-    %sum = "gpu.all_reduce"(%val) ({}) { op = "add" } : (f32) -> (f32)
+    %sum = gpu.all_reduce add %val {} : (f32) -> (f32)
     memref.store %sum, %dst[%tz, %ty, %tx] : memref<?x?x?xf32>
     gpu.terminator
   }
   call @print_memref_f32(%cast_dst) : (memref<*xf32>) -> ()
   return
 }
 
 func private @print_memref_f32(%ptr : memref<*xf32>)

mlir/test/Integration/GPU/CUDA/all-reduce-or.mlir

func @main() {
   memref.store %cst7, %data[%c1, %c3] : memref<2x6xi32>
   memref.store %cst10, %data[%c1, %c4] : memref<2x6xi32>
   memref.store %cst11, %data[%c1, %c5] : memref<2x6xi32>
 
   // OR
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c2, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %c6, %block_y = %c1, %block_z = %c1) {
     %val = memref.load %data[%bx, %tx] : memref<2x6xi32>
-    %reduced = "gpu.all_reduce"(%val) ({}) { op = "or" } : (i32) -> (i32)
+    %reduced = gpu.all_reduce or %val {} : (i32) -> (i32)
     memref.store %reduced, %sum[%bx] : memref<2xi32>
     gpu.terminator
   }
 
   call @print_memref_i32(%cast_sum) : (memref<*xi32>) -> ()
   // CHECK: [31, 15]
 
   return
 }
 
 func private @print_memref_i32(memref<*xi32>)
 

mlir/test/Integration/GPU/CUDA/all-reduce-region.mlir

func @main() {
   %one = arith.constant 1 : index
   %c0 = arith.constant 0 : index
   %sx = memref.dim %dst, %c0 : memref<?xf32>
   %cast_dst = memref.cast %dst : memref<?xf32> to memref<*xf32>
   gpu.host_register %cast_dst : memref<*xf32>
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %one, %grid_y = %one, %grid_z = %one)
              threads(%tx, %ty, %tz) in (%block_x = %sx, %block_y = %one, %block_z = %one) {
     %val = arith.index_cast %tx : index to i32
-    %xor = "gpu.all_reduce"(%val) ({
+    %xor = gpu.all_reduce %val {
     ^bb(%lhs : i32, %rhs : i32):
       %xor = arith.xori %lhs, %rhs : i32
       "gpu.yield"(%xor) : (i32) -> ()
-    }) : (i32) -> (i32)
+    } : (i32) -> (i32)
     %res = arith.sitofp %xor : i32 to f32
     memref.store %res, %dst[%tx] : memref<?xf32>
     gpu.terminator
   }
   call @print_memref_f32(%cast_dst) : (memref<*xf32>) -> ()
   return
 }
 
 func private @print_memref_f32(memref<*xf32>)

mlir/test/Integration/GPU/CUDA/all-reduce-xor.mlir

func @main() {
   memref.store %cst7, %data[%c1, %c3] : memref<2x6xi32>
   memref.store %cst10, %data[%c1, %c4] : memref<2x6xi32>
   memref.store %cst11, %data[%c1, %c5] : memref<2x6xi32>
 
   // XOR
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c2, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %c6, %block_y = %c1, %block_z = %c1) {
     %val = memref.load %data[%bx, %tx] : memref<2x6xi32>
-    %reduced = "gpu.all_reduce"(%val) ({}) { op = "xor" } : (i32) -> (i32)
+    %reduced = gpu.all_reduce xor %val {} : (i32) -> (i32)
     memref.store %reduced, %sum[%bx] : memref<2xi32>
     gpu.terminator
   }
 
   call @print_memref_i32(%cast_sum) : (memref<*xi32>) -> ()
   // CHECK: [31, 1]
 
   return
 }
 
 func private @print_memref_i32(memref<*xi32>)
 

mlir/test/Integration/GPU/CUDA/multiple-all-reduce.mlir

func @main() {
   memref.store %cst7, %data[%c1, %c3] : memref<2x6xf32>
   memref.store %cst10, %data[%c1, %c4] : memref<2x6xf32>
   memref.store %cst11, %data[%c1, %c5] : memref<2x6xf32>
 
   // ADD + MUL
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %c2, %grid_y = %c1, %grid_z = %c1)
              threads(%tx, %ty, %tz) in (%block_x = %c6, %block_y = %c1, %block_z = %c1) {
     %val = memref.load %data[%bx, %tx] : memref<2x6xf32>
-    %reduced0 = "gpu.all_reduce"(%val) ({}) { op = "add" } : (f32) -> (f32)
+    %reduced0 = gpu.all_reduce add %val {} : (f32) -> (f32)
     memref.store %reduced0, %sum[%bx] : memref<2xf32>
-    %reduced1 = "gpu.all_reduce"(%val) ({}) { op = "mul" } : (f32) -> (f32)
+    %reduced1 = gpu.all_reduce mul %val {} : (f32) -> (f32)
     memref.store %reduced1, %mul[%bx] : memref<2xf32>
     gpu.terminator
   }
 
   call @print_memref_f32(%cast_sum) : (memref<*xf32>) -> ()
   // CHECK: [31, 39]
 
   call @print_memref_f32(%cast_mul) : (memref<*xf32>) -> ()
   // CHECK: [0, 27720]
 
   return
 }
 
 func private @print_memref_f32(memref<*xf32>)

mlir/test/Integration/GPU/CUDA/shuffle.mlir

func @main() {
   %cast_dst = memref.cast %dst : memref<?xf32> to memref<*xf32>
   gpu.host_register %cast_dst : memref<*xf32>
   gpu.launch blocks(%bx, %by, %bz) in (%grid_x = %one, %grid_y = %one, %grid_z = %one)
              threads(%tx, %ty, %tz) in (%block_x = %sx, %block_y = %one, %block_z = %one) {
     %t0 = arith.index_cast %tx : index to i32
     %val = arith.sitofp %t0 : i32 to f32
     %width = arith.index_cast %block_x : index to i32
     %offset = arith.constant 4 : i32
-    %shfl, %valid = gpu.shuffle %val, %offset, %width xor : f32
+    %shfl, %valid = gpu.shuffle xor %val, %offset, %width : f32
     cond_br %valid, ^bb1(%shfl : f32), ^bb0
   ^bb0:
     %m1 = arith.constant -1.0 : f32
     br ^bb1(%m1 : f32)
   ^bb1(%value : f32):
     memref.store %value, %dst[%tx] : memref<?xf32>
     gpu.terminator
   }
   call @print_memref_f32(%cast_dst) : (memref<*xf32>) -> ()
   return
 }
 
 func private @print_memref_f32(%ptr : memref<*xf32>)

mlir/test/Integration/GPU/ROCM/printf.mlir

 // RUN: mlir-opt %s \
 // RUN:   -pass-pipeline='gpu.module(strip-debuginfo,convert-gpu-to-rocdl{index-bitwidth=32 runtime=HIP},gpu-to-hsaco{chip=%chip})' \
 // RUN:   -gpu-to-llvm \
 // RUN: | mlir-cpu-runner \
 // RUN:   --shared-libs=%linalg_test_lib_dir/libmlir_rocm_runtime%shlibext \
 // RUN:   --shared-libs=%linalg_test_lib_dir/libmlir_runner_utils%shlibext \
 // RUN:   --entry-point-result=void \
 // RUN: | FileCheck %s
 
 // CHECK: Hello from 0
 // CHECK: Hello from 1
 module attributes {gpu.container_module} {
     gpu.module @kernels {
         gpu.func @hello() kernel {
-            %0 = "gpu.thread_id"() {dimension="x"} : () -> (index)
+            %0 = gpu.thread_id x
             gpu.printf "Hello from %d\n" %0 : index
             gpu.return
         }
     }
 
     func @main() {
         %c2 = arith.constant 2 : index
         %c1 = arith.constant 1 : index
         gpu.launch_func @kernels::@hello
             blocks in (%c1, %c1, %c1)
             threads in (%c2, %c1, %c1)
         return
     }
 }

mlir/test/Target/LLVMIR/nvvmir.mlir

llvm.func @llvm_nvvm_barrier0() {
   nvvm.barrier0
   llvm.return
 }
 
 llvm.func @nvvm_shfl(
     %0 : i32, %1 : i32, %2 : i32,
     %3 : i32, %4 : f32) -> i32 {
   // CHECK: call i32 @llvm.nvvm.shfl.sync.bfly.i32(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %6 = nvvm.shfl.sync "bfly" %0, %3, %1, %2 : i32 -> i32
+  %6 = nvvm.shfl.sync bfly %0, %3, %1, %2 : i32 -> i32
   // CHECK: call float @llvm.nvvm.shfl.sync.bfly.f32(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %7 = nvvm.shfl.sync "bfly" %0, %4, %1, %2 : f32 -> f32
+  %7 = nvvm.shfl.sync bfly %0, %4, %1, %2 : f32 -> f32
   // CHECK: call i32 @llvm.nvvm.shfl.sync.up.i32(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %8 = nvvm.shfl.sync "up" %0, %3, %1, %2 : i32 -> i32
+  %8 = nvvm.shfl.sync up %0, %3, %1, %2 : i32 -> i32
   // CHECK: call float @llvm.nvvm.shfl.sync.up.f32(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %9 = nvvm.shfl.sync "up" %0, %4, %1, %2 : f32 -> f32
+  %9 = nvvm.shfl.sync up %0, %4, %1, %2 : f32 -> f32
   // CHECK: call i32 @llvm.nvvm.shfl.sync.down.i32(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %10 = nvvm.shfl.sync "down" %0, %3, %1, %2 : i32 -> i32
+  %10 = nvvm.shfl.sync down %0, %3, %1, %2 : i32 -> i32
   // CHECK: call float @llvm.nvvm.shfl.sync.down.f32(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %11 = nvvm.shfl.sync "down" %0, %4, %1, %2 : f32 -> f32
+  %11 = nvvm.shfl.sync down %0, %4, %1, %2 : f32 -> f32
   // CHECK: call i32 @llvm.nvvm.shfl.sync.idx.i32(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %12 = nvvm.shfl.sync "idx" %0, %3, %1, %2 : i32 -> i32
+  %12 = nvvm.shfl.sync idx %0, %3, %1, %2 : i32 -> i32
   // CHECK: call float @llvm.nvvm.shfl.sync.idx.f32(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %13 = nvvm.shfl.sync "idx" %0, %4, %1, %2 : f32 -> f32
+  %13 = nvvm.shfl.sync idx %0, %4, %1, %2 : f32 -> f32
   llvm.return %6 : i32
 }
 
 llvm.func @nvvm_shfl_pred(
     %0 : i32, %1 : i32, %2 : i32,
     %3 : i32, %4 : f32) -> !llvm.struct<(i32, i1)> {
   // CHECK: call { i32, i1 } @llvm.nvvm.shfl.sync.bfly.i32p(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %6 = nvvm.shfl.sync "bfly" %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
+  %6 = nvvm.shfl.sync bfly %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
   // CHECK: call { float, i1 } @llvm.nvvm.shfl.sync.bfly.f32p(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %7 = nvvm.shfl.sync "bfly" %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+  %7 = nvvm.shfl.sync bfly %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
   // CHECK: call { i32, i1 } @llvm.nvvm.shfl.sync.up.i32p(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %8 = nvvm.shfl.sync "up" %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
+  %8 = nvvm.shfl.sync up %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
   // CHECK: call { float, i1 } @llvm.nvvm.shfl.sync.up.f32p(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %9 = nvvm.shfl.sync "up" %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+  %9 = nvvm.shfl.sync up %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
   // CHECK: call { i32, i1 } @llvm.nvvm.shfl.sync.down.i32p(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %10 = nvvm.shfl.sync "down" %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
+  %10 = nvvm.shfl.sync down %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
   // CHECK: call { float, i1 } @llvm.nvvm.shfl.sync.down.f32p(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %11 = nvvm.shfl.sync "down" %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+  %11 = nvvm.shfl.sync down %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
   // CHECK: call { i32, i1 } @llvm.nvvm.shfl.sync.idx.i32p(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %12 = nvvm.shfl.sync "idx" %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
+  %12 = nvvm.shfl.sync idx %0, %3, %1, %2 {return_value_and_is_valid} : i32 -> !llvm.struct<(i32, i1)>
   // CHECK: call { float, i1 } @llvm.nvvm.shfl.sync.idx.f32p(i32 %{{.*}}, float %{{.*}}, i32 %{{.*}}, i32 %{{.*}})
-  %13 = nvvm.shfl.sync "idx" %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
+  %13 = nvvm.shfl.sync idx %0, %4, %1, %2 {return_value_and_is_valid} : f32 -> !llvm.struct<(f32, i1)>
   llvm.return %6 : !llvm.struct<(i32, i1)>
 }
 
 llvm.func @nvvm_vote(%0 : i32, %1 : i1) -> i32 {
   // CHECK: call i32 @llvm.nvvm.vote.ballot.sync(i32 %{{.*}}, i1 %{{.*}})
   %3 = nvvm.vote.ballot.sync %0, %1 : i32
   llvm.return %3 : i32
 }
 
 llvm.func @nvvm_mma(%a0 : vector<2xf16>, %a1 : vector<2xf16>,
                     %b0 : vector<2xf16>, %b1 : vector<2xf16>,
                     %c0 : f32, %c1 : f32, %c2 : f32, %c3 : f32,
                     %c4 : f32, %c5 : f32, %c6 : f32, %c7 : f32) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)> {
   // CHECK: call { float, float, float, float, float, float, float, float } @llvm.nvvm.mma.m8n8k4.row.col.f32.f32
   %0 = nvvm.mma.sync %a0, %a1, %b0, %b1, %c0, %c1, %c2, %c3, %c4, %c5, %c6, %c7 {alayout="row", blayout="col"} : (vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, f32, f32, f32, f32, f32, f32, f32, f32) -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
   llvm.return %0 : !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
 }
 
 // The test below checks the correct mapping of the nvvm.wmma.*.load.* op to the correct intrinsic
 // in the LLVM NVPTX backend.
 llvm.func @gpu_wmma_load_op(%arg0: !llvm.ptr<i32, 3>, %arg1: i32) {
   // CHECK: call { <2 x half>, <2 x half>, <2 x half>, <2 x half>, <2 x half>, <2 x half>, <2 x half>, <2 x half> } @llvm.nvvm.wmma.m16n16k16.load.a.row.stride.f16.p3i32(i32 addrspace(3)* %{{.*}}, i32 %{{.*}})
   %0 = nvvm.wmma.load %arg0, %arg1
-    {eltype = "f16", frag = "a", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+    {eltype = #nvvm.mma_type<f16>, frag = #nvvm.mma_frag<a>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (!llvm.ptr<i32, 3>) -> !llvm.struct<(vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>, vector<2xf16>)>
 
   llvm.return
 }
 
 // The test below checks the correct mapping of the nvvm.wmma.*.store.* op to the correct intrinsic
 // in the LLVM NVPTX backend.
 llvm.func @gpu_wmma_store_op(%arg0: !llvm.ptr<i32, 3>, %arg1: i32,
                             %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                             %arg4: vector<2 xf16>, %arg5: vector<2 x f16>) {
   // CHECK: call void @llvm.nvvm.wmma.m16n16k16.store.d.row.stride.f16.p3i32(i32 addrspace(3)* %{{.*}}, <2 x half> {{.*}}, <2 x half> %{{.*}}, <2 x half> %{{.*}}, <2 x half> %{{.*}}, i32 %{{.*}})
   nvvm.wmma.store %arg0, %arg1, %arg2, %arg3, %arg4, %arg5
-    {eltype = "f16", k = 16 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+    {eltype = #nvvm.mma_type<f16>, k = 16 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : !llvm.ptr<i32, 3>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>
   llvm.return
 }
 
 // The test below checks the correct mapping of the nvvm.wmma.*.mma.* op to the correct intrinsic
 // in the LLVM NVPTX backend.
 llvm.func @gpu_wmma_mma_op(%arg0: vector<2 x f16>, %arg1: vector<2 x f16>,
                         %arg2: vector<2 x f16>, %arg3: vector<2 x f16>,
                         %arg4: vector<2 x f16>, %arg5: vector<2 x f16>,
                         %arg6: vector<2 x f16>, %arg7: vector<2 x f16>,
                         %arg8: vector<2 x f16>, %arg9: vector<2 x f16>,
                         %arg10: vector<2 x f16>, %arg11: vector<2 x f16>,
                         %arg12: vector<2 x f16>, %arg13: vector<2 x f16>,
                         %arg14: vector<2 x f16>, %arg15: vector<2 x f16>,
                         %arg16: vector<2 x f16>, %arg17: vector<2 x f16>,
                         %arg18: vector<2 x f16>, %arg19: vector<2 x f16>) {
   // CHECK: call { <2 x half>, <2 x half>, <2 x half>, <2 x half> } @llvm.nvvm.wmma.m16n16k16.mma.row.row.f16.f16(<2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}}, <2 x half> {{.*}})
   %0 = nvvm.wmma.mma %arg0, %arg1, %arg2, %arg3, %arg4, %arg5, %arg6, %arg7, %arg8, %arg9, %arg10, %arg11, %arg12, %arg13, %arg14, %arg15, %arg16, %arg17, %arg18, %arg19
-    {eltypeA = "f16", eltypeB = "f16", k = 16 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<f16>, eltypeB = #nvvm.mma_type<f16>, k = 16 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>,
        vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>,
        vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>,
        vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>)
       -> !llvm.struct<(vector<2 x f16>, vector<2 x f16>, vector<2 x f16>, vector<2 x f16>)>
   llvm.return
 }
 
 llvm.func @nvvm_wmma_load_tf32(%arg0: !llvm.ptr<i32>, %arg1 : i32) {
   // CHECK: call { i32, i32, i32, i32 } @llvm.nvvm.wmma.m16n16k8.load.a.row.stride.tf32.p0i32(i32* %{{.*}}, i32 %{{.*}})
   %0 = nvvm.wmma.load %arg0, %arg1
-    {eltype = "tf32", frag = "a", k = 8 : i32, layout = "row", m = 16 : i32, n = 16 : i32}
+    {eltype = #nvvm.mma_type<tf32>, frag = #nvvm.mma_frag<a>, k = 8 : i32, layout = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (!llvm.ptr<i32>) -> !llvm.struct<(i32, i32, i32, i32)>
   llvm.return
 }
 
 llvm.func @nvvm_wmma_mma(%0 : i32, %1 : i32, %2 : i32, %3 : i32, %4 : i32, %5 : i32,
                     %6 : i32, %7 : i32, %8 : f32, %9 : f32, %10 : f32,
                     %11 : f32, %12 : f32, %13 : f32, %14 : f32, %15 : f32) {
   // CHECK: { float, float, float, float, float, float, float, float } @llvm.nvvm.wmma.m16n16k8.mma.row.row.tf32(i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, i32 %{{.*}}, float %{{.*}}, float %{{.*}}, float %{{.*}}, float %{{.*}}, float %{{.*}}, float %{{.*}}, float %{{.*}}, float %{{.*}})
   %r = nvvm.wmma.mma %0, %1, %2, %3, %4, %5, %6, %7, %8, %9, %10, %11, %12, %13, %14, %15
-    {eltypeA = "tf32", eltypeB = "f32", k = 8 : i32, layoutA = "row", layoutB = "row", m = 16 : i32, n = 16 : i32}
+    {eltypeA = #nvvm.mma_type<tf32>, eltypeB = #nvvm.mma_type<f32>, k = 8 : i32, layoutA = #nvvm.mma_layout<row>, layoutB = #nvvm.mma_layout<row>, m = 16 : i32, n = 16 : i32}
     : (i32, i32, i32, i32, i32, i32, i32, i32, f32, f32, f32, f32, f32, f32, f32, f32)
     -> !llvm.struct<(f32, f32, f32, f32, f32, f32, f32, f32)>
   llvm.return
 }
 
 llvm.func @cp_async(%arg0: !llvm.ptr<i8, 3>, %arg1: !llvm.ptr<i8, 1>) {
 // CHECK: call void @llvm.nvvm.cp.async.ca.shared.global.4(i8 addrspace(3)* %{{.*}}, i8 addrspace(1)* %{{.*}})
   nvvm.cp.async.shared.global %arg0, %arg1, 4

mlir/test/Target/LLVMIR/openmp-llvm.mlir

llvm.func @omp_ordered(%arg0 : i32, %arg1 : i32, %arg2 : i32, %arg3 : i64,
   }
 
   omp.wsloop (%arg7) : i32 = (%arg0) to (%arg1) step (%arg2) ordered(1) {
     // CHECK: [[TMP:%.*]] = getelementptr inbounds [1 x i64], [1 x i64]* [[ADDR]], i64 0, i64 0
     // CHECK: store i64 [[ARG0:%.*]], i64* [[TMP]], align 8
     // CHECK: [[TMP2:%.*]] = getelementptr inbounds [1 x i64], [1 x i64]* [[ADDR]], i64 0, i64 0
     // CHECK: [[OMP_THREAD2:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB3:[0-9]+]])
     // CHECK: call void @__kmpc_doacross_wait(%struct.ident_t* @[[GLOB3]], i32 [[OMP_THREAD2]], i64* [[TMP2]])
-    omp.ordered depend_type("dependsink") depend_vec(%arg3 : i64) {num_loops_val = 1 : i64}
+    omp.ordered depend_type(dependsink) depend_vec(%arg3 : i64) {num_loops_val = 1 : i64}
 
     // CHECK: [[TMP3:%.*]] = getelementptr inbounds [1 x i64], [1 x i64]* [[ADDR3]], i64 0, i64 0
     // CHECK: store i64 [[ARG0]], i64* [[TMP3]], align 8
     // CHECK: [[TMP4:%.*]] = getelementptr inbounds [1 x i64], [1 x i64]* [[ADDR3]], i64 0, i64 0
     // CHECK: [[OMP_THREAD4:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB5:[0-9]+]])
     // CHECK: call void @__kmpc_doacross_post(%struct.ident_t* @[[GLOB5]], i32 [[OMP_THREAD4]], i64* [[TMP4]])
-    omp.ordered depend_type("dependsource") depend_vec(%arg3 : i64) {num_loops_val = 1 : i64}
+    omp.ordered depend_type(dependsource) depend_vec(%arg3 : i64) {num_loops_val = 1 : i64}
 
     omp.yield
   }
 
   omp.wsloop (%arg7) : i32 = (%arg0) to (%arg1) step (%arg2) ordered(2) {
     // CHECK: [[TMP5:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR5]], i64 0, i64 0
     // CHECK: store i64 [[ARG0]], i64* [[TMP5]], align 8
     // CHECK: [[TMP6:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR5]], i64 0, i64 1
     // CHECK: store i64 [[ARG1:%.*]], i64* [[TMP6]], align 8
     // CHECK: [[TMP7:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR5]], i64 0, i64 0
     // CHECK: [[OMP_THREAD6:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB7:[0-9]+]])
     // CHECK: call void @__kmpc_doacross_wait(%struct.ident_t* @[[GLOB7]], i32 [[OMP_THREAD6]], i64* [[TMP7]])
     // CHECK: [[TMP8:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR7]], i64 0, i64 0
     // CHECK: store i64 [[ARG2:%.*]], i64* [[TMP8]], align 8
     // CHECK: [[TMP9:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR7]], i64 0, i64 1
     // CHECK: store i64 [[ARG3:%.*]], i64* [[TMP9]], align 8
     // CHECK: [[TMP10:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR7]], i64 0, i64 0
     // CHECK: [[OMP_THREAD8:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB7]])
     // CHECK: call void @__kmpc_doacross_wait(%struct.ident_t* @[[GLOB7]], i32 [[OMP_THREAD8]], i64* [[TMP10]])
-    omp.ordered depend_type("dependsink") depend_vec(%arg3, %arg4, %arg5, %arg6 : i64, i64, i64, i64) {num_loops_val = 2 : i64}
+    omp.ordered depend_type(dependsink) depend_vec(%arg3, %arg4, %arg5, %arg6 : i64, i64, i64, i64) {num_loops_val = 2 : i64}
 
     // CHECK: [[TMP11:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR9]], i64 0, i64 0
     // CHECK: store i64 [[ARG0]], i64* [[TMP11]], align 8
     // CHECK: [[TMP12:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR9]], i64 0, i64 1
     // CHECK: store i64 [[ARG1]], i64* [[TMP12]], align 8
     // CHECK: [[TMP13:%.*]] = getelementptr inbounds [2 x i64], [2 x i64]* [[ADDR9]], i64 0, i64 0
     // CHECK: [[OMP_THREAD10:%.*]] = call i32 @__kmpc_global_thread_num(%struct.ident_t* @[[GLOB9:[0-9]+]])
     // CHECK: call void @__kmpc_doacross_post(%struct.ident_t* @[[GLOB9]], i32 [[OMP_THREAD10]], i64* [[TMP13]])
-    omp.ordered depend_type("dependsource") depend_vec(%arg3, %arg4 : i64, i64) {num_loops_val = 2 : i64}
+    omp.ordered depend_type(dependsource) depend_vec(%arg3, %arg4 : i64, i64) {num_loops_val = 2 : i64}
 
     omp.yield
   }
 
   llvm.return
 }
 
 // -----

mlir/test/lib/Dialect/Linalg/TestLinalgDistribution.cpp

 #include "mlir/Dialect/Linalg/IR/Linalg.h"
 #include "mlir/Dialect/Linalg/Transforms/Transforms.h"
 #include "mlir/Pass/Pass.h"
 #include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 
 using namespace mlir;
 using namespace mlir::linalg;
 
-template <char dim>
+template <gpu::Dimension Dim>
 static linalg::ProcInfo getGpuBlockInfo(OpBuilder &b, Location loc) {
-  std::string d(1, dim);
-  StringAttr attr = b.getStringAttr(d);
-
   Type indexType = b.getIndexType();
-  ProcInfo procInfo = {b.create<gpu::BlockIdOp>(loc, indexType, attr),
-                       b.create<gpu::GridDimOp>(loc, indexType, attr)};
+  ProcInfo procInfo = {b.create<gpu::BlockIdOp>(loc, indexType, Dim),
+                       b.create<gpu::GridDimOp>(loc, indexType, Dim)};
   return procInfo;
 }
 
 static LinalgLoopDistributionOptions getDistributionOptions() {
   LinalgLoopDistributionOptions opts;
-  opts.procInfoMap.insert(std::make_pair("block_x", getGpuBlockInfo<'x'>));
-  opts.procInfoMap.insert(std::make_pair("block_y", getGpuBlockInfo<'y'>));
+  opts.procInfoMap.insert(
+      std::make_pair("block_x", getGpuBlockInfo<gpu::Dimension::x>));
+  opts.procInfoMap.insert(
+      std::make_pair("block_y", getGpuBlockInfo<gpu::Dimension::y>));
   return opts;
 }
 
 namespace {
 struct TestLinalgDistribution
     : public PassWrapper<TestLinalgDistribution, FunctionPass> {
   StringRef getArgument() const final { return "test-linalg-distribution"; }
   StringRef getDescription() const final { return "Test Linalg distribution."; }

mlir/test/lib/Dialect/Linalg/TestLinalgTransforms.cpp

patterns.add<LinalgPromotionPattern<MatmulOp>>(
       LinalgTransformationFilter(StringAttr::get(ctx, "PROMOTE")));
 }
 
 template <typename IdOp, typename NProcsOp>
 static SmallVector<ProcInfo, 2>
 getGpuProcIds(OpBuilder &b, Location loc, ArrayRef<Range> parallelLoopRanges) {
   size_t count = std::min<size_t>(3, parallelLoopRanges.size());
   SmallVector<ProcInfo, 2> procInfo(count);
-  const char *xyz[] = {"x", "y", "z"};
   Type indexType = b.getIndexType();
   for (unsigned i = 0; i < count; ++i) {
-    procInfo[count - 1 - i] = {
-        b.create<IdOp>(loc, indexType, b.getStringAttr(xyz[i])),
-        b.create<NProcsOp>(loc, indexType, b.getStringAttr(xyz[i]))};
+    gpu::Dimension dim = *gpu::symbolizeDimension(i);
+    procInfo[count - 1 - i] = {b.create<IdOp>(loc, indexType, dim),
+                               b.create<NProcsOp>(loc, indexType, dim)};
   }
   return procInfo;
 }
 
 static void fillTileAndDistributePatterns(MLIRContext *context,
                                           RewritePatternSet &patterns) {
   {
     LinalgLoopDistributionOptions cyclicNprocsEqNiters;

mlir/test/mlir-vulkan-runner/addf.mlir

 // RUN: mlir-vulkan-runner %s --shared-libs=%vulkan_wrapper_library_dir/libvulkan-runtime-wrappers%shlibext,%linalg_test_lib_dir/libmlir_runner_utils%shlibext --entry-point-result=void | FileCheck %s
 
 // CHECK: [3.3,  3.3,  3.3,  3.3,  3.3,  3.3,  3.3,  3.3]
 module attributes {
   gpu.container_module,
   spv.target_env = #spv.target_env<
     #spv.vce<v1.0, [Shader], [SPV_KHR_storage_buffer_storage_class]>, {}>
 } {
   gpu.module @kernels {
     gpu.func @kernel_add(%arg0 : memref<8xf32>, %arg1 : memref<8xf32>, %arg2 : memref<8xf32>)
       kernel attributes { spv.entry_point_abi = {local_size = dense<[1, 1, 1]>: vector<3xi32> }} {
-      %0 = "gpu.block_id"() {dimension = "x"} : () -> index
+      %0 = gpu.block_id x
       %1 = memref.load %arg0[%0] : memref<8xf32>
       %2 = memref.load %arg1[%0] : memref<8xf32>
       %3 = arith.addf %1, %2 : f32
       memref.store %3, %arg2[%0] : memref<8xf32>
       gpu.return
     }
   }
 

mlir/test/mlir-vulkan-runner/addi.mlir

 // RUN: mlir-vulkan-runner %s --shared-libs=%vulkan_wrapper_library_dir/libvulkan-runtime-wrappers%shlibext,%linalg_test_lib_dir/libmlir_runner_utils%shlibext --entry-point-result=void | FileCheck %s
 
 // CHECK-COUNT-64: [3, 3, 3, 3, 3, 3, 3, 3]
 module attributes {
   gpu.container_module,
   spv.target_env = #spv.target_env<
     #spv.vce<v1.0, [Shader], [SPV_KHR_storage_buffer_storage_class]>, {}>
 } {
   gpu.module @kernels {
     gpu.func @kernel_addi(%arg0 : memref<8xi32>, %arg1 : memref<8x8xi32>, %arg2 : memref<8x8x8xi32>)
       kernel attributes { spv.entry_point_abi = {local_size = dense<[1, 1, 1]>: vector<3xi32>}} {
-      %x = "gpu.block_id"() {dimension = "x"} : () -> index
-      %y = "gpu.block_id"() {dimension = "y"} : () -> index
-      %z = "gpu.block_id"() {dimension = "z"} : () -> index
+      %x = gpu.block_id x
+      %y = gpu.block_id y
+      %z = gpu.block_id z
       %0 = memref.load %arg0[%x] : memref<8xi32>
       %1 = memref.load %arg1[%y, %x] : memref<8x8xi32>
       %2 = arith.addi %0, %1 : i32
       memref.store %2, %arg2[%z, %y, %x] : memref<8x8x8xi32>
       gpu.return
     }
   }
 

mlir/test/mlir-vulkan-runner/addi8.mlir

 // RUN: mlir-vulkan-runner %s --shared-libs=%vulkan_wrapper_library_dir/libvulkan-runtime-wrappers%shlibext,%linalg_test_lib_dir/libmlir_runner_utils%shlibext --entry-point-result=void | FileCheck %s
 
 // CHECK-COUNT-64: [3, 3, 3, 3, 3, 3, 3, 3]
 module attributes {
   gpu.container_module,
   spv.target_env = #spv.target_env<
     #spv.vce<v1.0, [Shader], [SPV_KHR_storage_buffer_storage_class, SPV_KHR_8bit_storage]>, {}>
 } {
   gpu.module @kernels {
     gpu.func @kernel_addi(%arg0 : memref<8xi8>, %arg1 : memref<8x8xi8>, %arg2 : memref<8x8x8xi32>)
       kernel attributes { spv.entry_point_abi = {local_size = dense<[1, 1, 1]>: vector<3xi32>}} {
-      %x = "gpu.block_id"() {dimension = "x"} : () -> index
-      %y = "gpu.block_id"() {dimension = "y"} : () -> index
-      %z = "gpu.block_id"() {dimension = "z"} : () -> index
+      %x = gpu.block_id x
+      %y = gpu.block_id y
+      %z = gpu.block_id z
       %0 = memref.load %arg0[%x] : memref<8xi8>
       %1 = memref.load %arg1[%y, %x] : memref<8x8xi8>
       %2 = arith.addi %0, %1 : i8
       %3 = arith.extui %2 : i8 to i32
       memref.store %3, %arg2[%z, %y, %x] : memref<8x8x8xi32>
       gpu.return
     }
   }

mlir/test/mlir-vulkan-runner/mulf.mlir

 // RUN: mlir-vulkan-runner %s --shared-libs=%vulkan_wrapper_library_dir/libvulkan-runtime-wrappers%shlibext,%linalg_test_lib_dir/libmlir_runner_utils%shlibext --entry-point-result=void | FileCheck %s
 
 // CHECK-COUNT-4: [6, 6, 6, 6]
 module attributes {
   gpu.container_module,
   spv.target_env = #spv.target_env<
     #spv.vce<v1.0, [Shader], [SPV_KHR_storage_buffer_storage_class]>, {}>
 } {
   gpu.module @kernels {
     gpu.func @kernel_mul(%arg0 : memref<4x4xf32>, %arg1 : memref<4x4xf32>, %arg2 : memref<4x4xf32>)
       kernel attributes { spv.entry_point_abi = {local_size = dense<[1, 1, 1]>: vector<3xi32> }} {
-      %x = "gpu.block_id"() {dimension = "x"} : () -> index
-      %y = "gpu.block_id"() {dimension = "y"} : () -> index
+      %x = gpu.block_id x
+      %y = gpu.block_id y
       %1 = memref.load %arg0[%x, %y] : memref<4x4xf32>
       %2 = memref.load %arg1[%x, %y] : memref<4x4xf32>
       %3 = arith.mulf %1, %2 : f32
       memref.store %3, %arg2[%x, %y] : memref<4x4xf32>
       gpu.return
     }
   }
 

mlir/test/mlir-vulkan-runner/subf.mlir

 // RUN: mlir-vulkan-runner %s --shared-libs=%vulkan_wrapper_library_dir/libvulkan-runtime-wrappers%shlibext,%linalg_test_lib_dir/libmlir_runner_utils%shlibext --entry-point-result=void | FileCheck %s
 
 // CHECK-COUNT-32: [2.2, 2.2, 2.2, 2.2]
 module attributes {
   gpu.container_module,
   spv.target_env = #spv.target_env<
     #spv.vce<v1.0, [Shader], [SPV_KHR_storage_buffer_storage_class]>, {}>
 } {
   gpu.module @kernels {
     gpu.func @kernel_sub(%arg0 : memref<8x4x4xf32>, %arg1 : memref<4x4xf32>, %arg2 : memref<8x4x4xf32>)
       kernel attributes { spv.entry_point_abi = {local_size = dense<[1, 1, 1]>: vector<3xi32> }} {
-      %x = "gpu.block_id"() {dimension = "x"} : () -> index
-      %y = "gpu.block_id"() {dimension = "y"} : () -> index
-      %z = "gpu.block_id"() {dimension = "z"} : () -> index
+      %x = gpu.block_id x
+      %y = gpu.block_id y
+      %z = gpu.block_id z
       %1 = memref.load %arg0[%x, %y, %z] : memref<8x4x4xf32>
       %2 = memref.load %arg1[%y, %z] : memref<4x4xf32>
       %3 = arith.subf %1, %2 : f32
       memref.store %3, %arg2[%x, %y, %z] : memref<8x4x4xf32>
       gpu.return
     }
   }
 

mlir/test/mlir-vulkan-runner/time.mlir

 // RUN: mlir-vulkan-runner %s --shared-libs=%vulkan_wrapper_library_dir/libvulkan-runtime-wrappers%shlibext,%linalg_test_lib_dir/libmlir_runner_utils%shlibext --entry-point-result=void | FileCheck %s
 
 // CHECK: Compute shader execution time
 // CHECK: Command buffer submit time
 // CHECK: Wait idle time
 
 module attributes {
   gpu.container_module,
   spv.target_env = #spv.target_env<
     #spv.vce<v1.0, [Shader], [SPV_KHR_storage_buffer_storage_class]>, {}>
 } {
   gpu.module @kernels {
     gpu.func @kernel_add(%arg0 : memref<16384xf32>, %arg1 : memref<16384xf32>, %arg2 : memref<16384xf32>)
       kernel attributes { spv.entry_point_abi = {local_size = dense<[128, 1, 1]>: vector<3xi32> }} {
-      %bid = "gpu.block_id"() {dimension = "x"} : () -> index
-      %tid = "gpu.thread_id"() {dimension = "x"} : () -> index
+      %bid = gpu.block_id x
+      %tid = gpu.thread_id x
       %cst = arith.constant 128 : index
       %b = arith.muli %bid, %cst : index
       %0 = arith.addi %b, %tid : index
       %1 = memref.load %arg0[%0] : memref<16384xf32>
       %2 = memref.load %arg1[%0] : memref<16384xf32>
       %3 = arith.addf %1, %2 : f32
       memref.store %3, %arg2[%0] : memref<16384xf32>
       gpu.return

mlir/tools/mlir-tblgen/AttrOrTypeDefGen.cpp

::mlir::Attribute {0}::parseAttribute(::mlir::DialectAsmParser &parser,
   if (::mlir::failed(parser.parseKeyword(&attrTag)))
     return {{};
   {{
     ::mlir::Attribute attr;
     auto parseResult = generatedAttributeParser(parser, attrTag, type, attr);
     if (parseResult.hasValue())
       return attr;
   }
-  parser.emitError(typeLoc) << "unknown  attribute `"
+  parser.emitError(typeLoc) << "unknown attribute `"
       << attrTag << "` in dialect `" << getNamespace() << "`";
   return {{};
 }
 /// Print an attribute registered to this dialect.
 void {0}::printAttribute(::mlir::Attribute attr,
                          ::mlir::DialectAsmPrinter &printer) const {{
   if (::mlir::succeeded(generatedAttributePrinter(attr, printer)))
     return;

mlir/tools/mlir-tblgen/OpFormatGen.cpp

private:
     /// If the type is resolved based upon another operand or result, this is
     /// the variable or the attribute that this type is resolved to.
     ConstArgument resolver;
     /// If the type is resolved based upon another operand or result, this is
     /// a transformer to apply to the variable when resolving.
     Optional<StringRef> variableTransformer;
   };
 
+  /// The context in which an element is generated.
+  enum class GenContext {
+    /// The element is generated at the top-level or with the same behaviour.
+    Normal,
+    /// The element is generated inside an optional group.
+    Optional
+  };
+
   OperationFormat(const Operator &op)
       : allOperands(false), allOperandTypes(false), allResultTypes(false),
         infersResultTypes(false) {
     operandTypes.resize(op.getNumOperands(), TypeResolution());
     resultTypes.resize(op.getNumResults(), TypeResolution());
 
     hasImplicitTermTrait = llvm::any_of(op.getTraits(), [](const Trait &trait) {
       return trait.getDef().isSubClassOf("SingleBlockImplicitTerminator");
     });
 
     hasSingleBlockTrait =
         hasImplicitTermTrait || op.getTrait("::mlir::OpTrait::SingleBlock");
   }
 
   /// Generate the operation parser from this format.
   void genParser(Operator &op, OpClass &opClass);
   /// Generate the parser code for a specific format element.
   void genElementParser(Element *element, MethodBody &body,
-                        FmtContext &attrTypeCtx);
+                        FmtContext &attrTypeCtx,
+                        GenContext genCtx = GenContext::Normal);
   /// Generate the C++ to resolve the types of operands and results during
   /// parsing.
   void genParserTypeResolution(Operator &op, MethodBody &body);
   /// Generate the C++ to resolve the types of the operands during parsing.
   void genParserOperandTypeResolution(
       Operator &op, MethodBody &body,
       function_ref<void(TypeResolution &, StringRef)> emitTypeResolver);
   /// Generate the C++ to resolve regions during parsing.
void OperationFormat::genParser(Operator &op, OpClass &opClass) {
   genParserSuccessorResolution(op, body);
   genParserVariadicSegmentResolution(op, body);
   genParserTypeResolution(op, body);
 
   body << "  return ::mlir::success();\n";
 }
 
 void OperationFormat::genElementParser(Element *element, MethodBody &body,
-                                       FmtContext &attrTypeCtx) {
+                                       FmtContext &attrTypeCtx,
+                                       GenContext genCtx) {
   /// Optional Group.
   if (auto *optional = dyn_cast<OptionalElement>(element)) {
     auto elements = llvm::drop_begin(optional->getThenElements(),
                                      optional->getParseStart());
 
     // Generate a special optional parser for the first element to gate the
     // parsing of the rest of the elements.
     Element *firstElement = &*elements.begin();
if (auto *optional = dyn_cast<OptionalElement>(element)) {
     if (anchorAttr && anchorAttr != firstElement && anchorAttr->isUnitAttr()) {
       elidedAnchorElement = anchorAttr;
 
       // Add the anchor unit attribute to the operation state.
       body << "    result.addAttribute(\"" << anchorAttr->getVar()->name
            << "\", parser.getBuilder().getUnitAttr());\n";
     }
 
-    // Generate the rest of the elements normally.
+    // Generate the rest of the elements inside an optional group. Elements in
+    // an optional group after the guard are parsed as required.
     for (Element &childElement : llvm::drop_begin(elements, 1)) {
-      if (&childElement != elidedAnchorElement)
-        genElementParser(&childElement, body, attrTypeCtx);
+      if (&childElement != elidedAnchorElement) {
+        genElementParser(&childElement, body, attrTypeCtx,
+                         GenContext::Optional);
+      }
     }
     body << "  }";
 
     // Generate the else elements.
     auto elseElements = optional->getElseElements();
     if (!elseElements.empty()) {
       body << " else {\n";
       for (Element &childElement : elseElements)
if (auto *optional = dyn_cast<OptionalElement>(element)) {
     // attribute.
     std::string attrTypeStr;
     if (Optional<StringRef> typeBuilder = attr->getTypeBuilder()) {
       llvm::raw_string_ostream os(attrTypeStr);
       os << tgfmt(*typeBuilder, &attrTypeCtx);
     } else {
       attrTypeStr = "::mlir::Type{}";
     }
-    if (var->attr.isOptional()) {
+    if (genCtx == GenContext::Normal && var->attr.isOptional()) {
       body << formatv(optionalAttrParserCode, var->name, attrTypeStr);
     } else {
       if (attr->shouldBeQualified() ||
           var->attr.getStorageType() == "::mlir::Attribute")
         body << formatv(genericAttrParserCode, var->name, attrTypeStr);
       else
         body << formatv(attrParserCode, var->name, attrTypeStr);
     }
if (shouldFormatSymbolNameAttr(var)) {
            << "Attr().getValue());\n";
       return;
     }
 
     // Elide the attribute type if it is buildable.
     if (attr->getTypeBuilder())
       body << "  _odsPrinter.printAttributeWithoutType("
            << op.getGetterName(var->name) << "Attr());\n";
-    else if (var->attr.isOptional())
-      body << "_odsPrinter.printAttribute(" << op.getGetterName(var->name)
-           << "Attr());\n";
     else if (attr->shouldBeQualified() ||
              var->attr.getStorageType() == "::mlir::Attribute")
       body << "  _odsPrinter.printAttribute(" << op.getGetterName(var->name)
            << "Attr());\n";
     else
       body << "_odsPrinter.printStrippedAttrOrType("
            << op.getGetterName(var->name) << "Attr());\n";
   } else if (auto *operand = dyn_cast<OperandVariable>(element)) {

utils/bazel/llvm-project-overlay/mlir/BUILD.bazel

tbl_outs = [
         (
             ["-gen-enum-decls"],
             "include/mlir/Dialect/GPU/GPUOpsEnums.h.inc",
         ),
         (
             ["-gen-enum-defs"],
             "include/mlir/Dialect/GPU/GPUOpsEnums.cpp.inc",
         ),
+        (
+            ["-gen-attrdef-decls"],
+            "include/mlir/Dialect/GPU/GPUOpsAttributes.h.inc",
+        ),
+        (
+            ["-gen-attrdef-defs"],
+            "include/mlir/Dialect/GPU/GPUOpsAttributes.cpp.inc",
+        ),
     ],
     tblgen = ":mlir-tblgen",
     td_file = "include/mlir/Dialect/GPU/GPUOps.td",
     deps = [
         ":DLTIDialectTdFiles",
         ":GPUOpsTdFiles",
     ],
 )
tbl_outs = [
         (
             ["-gen-enum-decls"],
             "include/mlir/Dialect/LLVMIR/NVVMOpsEnums.h.inc",
         ),
         (
             ["-gen-enum-defs"],
             "include/mlir/Dialect/LLVMIR/NVVMOpsEnums.cpp.inc",
         ),
+        (
+            [
+                "-gen-attrdef-decls",
+                "-attrdefs-dialect=nvvm",
+            ],
+            "include/mlir/Dialect/LLVMIR/NVVMOpsAttributes.h.inc",
+        ),
+        (
+            [
+                "-gen-attrdef-defs",
+                "-attrdefs-dialect=nvvm",
+            ],
+            "include/mlir/Dialect/LLVMIR/NVVMOpsAttributes.cpp.inc",
+        ),
     ],
     tblgen = ":mlir-tblgen",
     td_file = "include/mlir/Dialect/LLVMIR/NVVMOps.td",
     deps = [":NVVMOpsTdFiles"],
 )
 
 gentbl_cc_library(
     name = "NVVMConversionIncGen",
tbl_outs = [
             ["-gen-enum-decls"],
             "include/mlir/Dialect/OpenACC/OpenACCOpsEnums.h.inc",
         ),
         (
             ["-gen-enum-defs"],
             "include/mlir/Dialect/OpenACC/OpenACCOpsEnums.cpp.inc",
         ),
         (
+            [
+                "-gen-attrdef-decls",
+                "-attrdefs-dialect=acc",
+            ],
+            "include/mlir/Dialect/OpenACC/OpenACCOpsAttributes.h.inc",
+        ),
+        (
+            [
+                "-gen-attrdef-defs",
+                "-attrdefs-dialect=acc",
+            ],
+            "include/mlir/Dialect/OpenACC/OpenACCOpsAttributes.cpp.inc",
+        ),
+        (
             ["-gen-op-doc"],
             "g3doc/Dialects/OpenACC/OpenACCOps.md",
         ),
     ],
     tblgen = ":mlir-tblgen",
     td_file = "include/mlir/Dialect/OpenACC/OpenACCOps.td",
     deps = [":OpenAccOpsTdFiles"],
 )
tbl_outs = [
             ["-gen-type-interface-decls"],
             "include/mlir/Dialect/OpenMP/OpenMPTypeInterfaces.h.inc",
         ),
         (
             ["-gen-type-interface-defs"],
             "include/mlir/Dialect/OpenMP/OpenMPTypeInterfaces.cpp.inc",
         ),
         (
+            [
+                "-gen-attrdef-decls",
+                "-attrdefs-dialect=omp",
+            ],
+            "include/mlir/Dialect/OpenMP/OpenMPOpsAttributes.h.inc",
+        ),
+        (
+            [
+                "-gen-attrdef-defs",
+                "-attrdefs-dialect=omp",
+            ],
+            "include/mlir/Dialect/OpenMP/OpenMPOpsAttributes.cpp.inc",
+        ),
+        (
             ["-gen-op-doc"],
             "g3doc/Dialects/OpenMP/OpenMPOps.md",
         ),
     ],
     tblgen = ":mlir-tblgen",
     td_file = "include/mlir/Dialect/OpenMP/OpenMPOps.td",
     deps = [":OpenMPOpsTdFiles"],
 )